Urban Stream Corridors and Forest Patches—The Connections: A Case Study of Bloomington, IN

Abstract

Streams and forests are ecosystems connected through hydrology, but few studies have looked at the connectivity between streams and forests in the context of urban development. City-made decisions affect connections between streams and forests by isolating both ecosystems. Streams are often channelized or buried to increase potential development areas. Forests often become fragmented and may be removed unless protected. Historical choices in land usage affect the sites and sizes of current urban streams, forests, and development. This affects the distribution of impervious surfaces, which separates streams from forests. Despite these barriers to stream/forest interactions, cities can experience stream/forest connectivity. Seven Bloomington watersheds are ranked on their proportions of buried streams, channelized streams, forested hydrology, forested streams, urban forest patch cover, and impervious surface cover, along with the historical presence of urbanization. Watersheds demonstrate stream/forest connectivity, with five watersheds containing 50% or greater forested stream segments. Bloomington canopy cover reduces stormwater runoff by approximately 127 kiloliters per year. These forested areas reduce flooding, reduce nutrient loading, and reduce stream conditions associated with urban stream syndrome. Understanding urban stream/forest connectivity can improve green infrastructure design and green space design, which improve urban resilience and better connect residents to the environment.

1. Introduction

Urban areas usually develop around streams. These channels provide stormwater drainage for communities and drinking water, both of which are necessary for environmental sustainability and human health in urban areas [1,2]. Urban development often impairs stream function by increasing nutrient and contaminant loads and altering stream flows, which usually causes greater variation in water volume. These changes typically alter channel morphologies and reduce species richness in stream channels [3,4]. Additionally, urban development occurs concurrently with new impervious surface cover (ISC). Development projects frequently remove or reduce the size of Urban Forest Patches (UFPs). UFPs are one patch type within a matrix of patches that also includes line corridors, networks, and stream corridors, and that in an urban setting is overwhelmingly modified and altered by humans [3,5].

UFPs are well-known for their function of improving air quality by removing air pollution, sequestering carbon, and regulating local climates. They also provide food production and recreational opportunities, remove pollutants from streams, and offer aesthetic value [6,7,8]. Urban development can reduce the ability of UFPs to provide ecosystem services through exposure to higher concentrations of contaminants and reduced species richness [9,10]. For instance, it is expected that atmospheric nitrogen deposition has a negative effect on urban forest growth. Once a tolerance threshold is crossed, forests with high atmospheric nitrogen deposition could have an increase in the number of invasive species and a negative impact on soil microbial diversity. In addition to atmospheric nitrogen deposition, soil nitrogen concentrations are consistently higher in urban forests than in outlying forested lands [11,12]. There have been multiple studies that explored the effects of higher CO₂ concentrations on tree growth (and subsequent increases in the ability to sequester carbon) at the whole-tree level. The effects of elevated CO₂ levels cause accelerated growth in seedlings but are variable in mature trees. This means predictions remain uncertain as to whether continued high CO₂ emissions can be compensated for by accelerated tree growth [3,11,12,13].

When connected to a stream, UFPs can improve the stream’s ecosystem dynamics and regulate flooding via avoided runoff, interception, and transpiration [14,15]. Connectivity between urban streams and UFPs depends on anthropogenic development near the stream and the proximity of UFPs to streams via spatial distribution, surface flow, and underground water tables [3,5,15]. For urban streams, stream-to-UFP connectivity decreases when the corridors are channelized or buried. For UFPs, being further away from the stream naturally decreases stream-to-UFP connectivity. However, stream-to-UFP connectivity also decreases when impervious surfaces exist between streams and UFPs. These relationships between streams and UFPs change over time [3,9]. Currently, many cities are expanding and developing in nearby rural areas [16]. Few studies to this point have investigated relationships between UFPs and forested stream ecosystems. To date, most work on stream-to-UFP connectivity and spatial distribution has studied rural systems, UFPs’ ability to provide ecosystem services, or habitat connectivity. This paper considers urban development over time as an application to Foreman’s landscape ecology theory of patches as chronic and expanding disturbances to forest patches, stream channels, and their connectivity in a landscape matrix [5]. Knowing how cities grow and develop is important to understanding how their streams and UFPs arrived at their current conditions and placement [3,5,14,17,18,19,20,21]

Within this paper are several technical related to urban grey and green infrastructure. These are defined below for clarity (

Table 1. Definitions of terms used in this paper.

Term	Definition
Channelization	Anthropogenic modifications to an urban stream that control erosion, drain wetlands, and reduce the retention time of water within the basin. Common technologies used to channelize stream corridors include concrete banks and rip-rap [22].
Buried Stream Segment	A section of a stream that has been funneled through a buried pipe to channelize the stream and increase constructible surface area [23].
Channelized Stream Segment	A section of a stream that has been modified to control erosion and reduce retention time. Includes sections of streams that have cement banks or have been lined with rip-rap [23].
Urban Forest Patch	A self-established ecosystem within city limits that includes enough plants to establish a canopy and an understory. Understories consist of a shrub and a ground cover layer [24].
Connectivity	The ability of surface or underground waters to move between forest and basin [25].
Impervious Surface Cover	Land cover that reduces the ability of water to infiltrate the ground. ISC includes buildings, roads, sidewalks, driveways, and other manufactured surfaces [26].

).

Research Questions

1.

(a) Where and when have Bloomington streams been channelized and/or buried over time?

 

(b) How have these modifications altered channel connectivity to nearby UFPs?

2.

(a) Where and when have Bloomington UFPs changed over time?

 

(b) How have these changes in the UFP area altered their impact on streams?

2. Materials and Methods

2.1. Study Area

Bloomington, Indiana, was chosen as the study site to observe the connections between streams and UFPs. The city contains about 79,000 residents and is located on rolling hills in southern Indiana [27,28]. The central and eastern parts of Bloomington have flatter topography. The west side of Bloomington exhibits notable karst topography, which is known to contain a number of sinking streams [29,30]. Bloomington was selected as the study site due to this community’s higher vulnerability to flood and drought events compared to similar-sized communities in similar ecoregions. Bloomington has a history of experiencing both major floods and major droughts that compromise the lives of residents [31,32,33]. In the case of sustainable city management, it is important to have a good understanding of these stream systems because the community’s entire streamflow moves out of the city core.

Bloomington has a land area of 60.17 square kilometers, with lakes and reservoirs covering an additional 0.48 square kilometers [34]. Bloomington’s water is dispersed throughout ten watersheds that flow out of the city. These watersheds are parts of two different, larger watersheds. The Lower White River covers the northern third of Bloomington, and the Lower East Fork White River covers the lower two-thirds of the city. Bloomington is a unique city hydrologically because its urban core sits on a divide between two larger watersheds. This means that all stream corridors in Bloomington flow away from the city [35,36].

Table 2. Subwatersheds within the larger watersheds in Bloomington.

Larger Watershed	Watersheds
Lower White River	Cascade Creek, Griffy Creek, and Stout Creek [35,36].
Lower East Fork White River	Clear Creek, Jackson Creek, East Fork Jackson Creek (EFJC), West Fork Clear Creek (WFCC), Stephens Creek (not included), Sinking Creek (not included), and Leonard Springs (not included) [35,36].

shows the ten watersheds that belong to each of the larger watersheds. Three watersheds were excluded from this study due to a lack of stream segments within Bloomington city limits or poorly defined watershed boundaries.

2.2. Spatial Data Sources

Geographic Information Systems (GIS) imagery, analyzed on ArcGIS Pro version 2.9, is used to observe and measure most of the metrics of this study. Three layers were downloaded from the Bloomington City GIS portal. First is the Bloomington 2019 city boundary. All maps produced for this study are clipped to this layer, which continues to be the current boundary as of 2022 [37]. Second is the Natural Drainage Basins layer, which shows subwatersheds within Bloomington. This layer has been modified using the 2019 Storm Outfalls Map to show the seven subwatershed boundaries within the study site [36,38]. The final source from the Bloomington City GIS portal is the Creeks and Streams layer. This layer maps all the surface streams. Comparing the Creeks and Streams layer to the Local Resolution Hydrology and GIS World imagery layers showed which streams are surface or sinking streams [39,40,41].

The Local Resolution Hydrology layer was downloaded from IndianaMap and contains all channels that are at least 12 m wide. This layer includes buried segments and channelized segments and provides an indication of buried stream segments and channelized stream segments [41]. Comparing this layer to the Creek and Streams layer, the 2019 Storm Outfalls Map, and GIS World imagery provided data on which channel segments are buried or channelized [36,40,41]. Additionally, segments less than 12 m wide were added if watersheds contained disconnected segments. Buried and channelized stream segments were compared to respective hydrology sections to obtain proportions of each watershed that are buried or channelized [40].

Indiana University Archives provided Bloomington aerial imagery for 1939 and 1967. The 1939 aerial imagery was chosen because it is the oldest aerial imagery for Bloomington and has enough image resolution to view land cover types. The 1967 aerial imagery was selected because it is the only other year to have a complete image of Bloomington with resolution to view land cover types. These images were taken as screenshots and then georeferenced in ArcGIS Pro. Maps show the locations and sizes of UFPs and streams for their respective years. Additionally, these maps show how Bloomington developed and expanded between 1939 and 1967 [42].

The National Land Cover Database (NLCD) provided data for the UFP and ISC layers. The UFP imagery was constructed using the NLCD 2016 canopy cover layer at 30-meter resolution. This is the most recent publicly available canopy cover imagery [43]. UFP imagery data was used to collect the percent forest cover for each watershed. Additionally, the Creeks and Streams and Local Resolution Hydrology layers were clipped to the UFP layer for each watershed to find the percent of forested streams and forested hydrology, respectively. Attribute tables were compiled for each watershed’s UFP areas, forested stream lengths, and forested hydrology lengths. Total UFP areas, forested stream lengths, and forested hydrology lengths were calculated and compared to total watershed areas, stream lengths, and hydrology lengths, respectively. For ISC, both 2001 and 2019 images are used. The NLCD 2001 ISC layer was the first ISC layer developed for the United States. The NLCD 2019 ISC is the most recent map that has been produced [44]. The average ISC percent was compiled using GIS for each watershed for both years.

Once all layers were collected and modified, maps were constructed using the Bloomington city boundary and Natural Drainage Basins layer and clipping all layers to current city limits and major watersheds [34,35]. Buried Hydrology and Channelized Hydrology layers were created by comparing Local Hydrology to the 2019 Storm Outfalls map, World imagery, and Creeks and Streams layer. Comparing these layers determined which channel segments are buried or flow through concrete-lined channels or rip-rap [36,39,40,41]. Bloomington Creeks and Streams and hydrology maps were created to determine where streams are located and where they have been modified. The UFP layer was added to the Creeks and Streams layer and the Local Resolution Hydrology layer to show the placement of current UFPs compared to surface streams and total hydrology [39,41,43]. The historical imagery for Bloomington was mapped by adding aerial imagery to the Bloomington city boundary layer and the Natural Drainage Basin layer. Comparing the 1939 to the 1967 aerial imagery indicates how Bloomington developed during this time [37,38,42]. Maps for ISC were constructed by adding ISC layers to the Bloomington city boundary layer and the Natural Drainage Basin layer. These maps show urban modifications and provide additional historical data [37,38,44]. Calculations were collected by clipping layers to each major watershed and accessing layer data attribute tables for stream and UFP proportions. These calculations were completed using Microsoft Office 365 Excel. The Excel spreadsheet is Table S1. In aerial imagery, urbanization was visually determined. Percent ISC was collected from average raster data values for each watershed. Finally, all calculations are used to rank each watershed, with less modified or urbanized watersheds receiving lower scores, indicating less anthropogenic influence.

2.3. Visual Inspection

A visual inspection of Bloomington UFPs was performed between March and June 2022 to ground-truth the data seen on the NLCD database, with the intent to investigate any general changes in the UFPs since 2016. Methods for this inspection included driving out to each site, visiting all UFPs, and confirming the sizes, shapes, and percent canopy cover of UFPs, the general overstory species assemblage, and the presence of streams. Observed tree species and assemblages were recorded at each site. Additionally, this inspection provided insight on sites that have natural streams, buried streams, or channelized streams. Channelized streams include surface channels lined by rip-rap or cement banks. This survey of urban forested patches added to the GIS imagery by refining the Local Resolution Hydrology, buried stream imagery, and UFP imagery.

2.4. Estimate of UFP Ecosystem Services

In addition to GIS imagery, estimates of Bloomington’s UFP ecosystem services were computed using i-Tree Canopy. i-Tree Canopy estimates the amounts of air pollutants, carbon sequestration, and mitigated storm runoff by calculating the percentage of canopy cover in an area. For this study, the Bloomington city limits were accessed from the US Census Places within the i-Tree Canopy software version 4. The i-Tree Canopy US Census Places map shows the current areas of towns and cities, which provide the same political boundaries as the city boundary used in the GIS imagery. The program selected random points around the city, and these points are labeled as either “tree/shrub” or “not tree/shrub”. For this estimation, 1000 random points were placed within Bloomington’s boundaries and categorized. Points had to be placed within a canopy cover that has an area of one acre or more to be designated as a tree/shrub. This designation is used to collect ecosystem service estimates more accurately for UFPs. i-Tree Canopy provided estimates for hydrological benefits. All evaluation estimates have been rounded to the nearest square kilometer [45].

3. Results

3.1. Channelization and Burial of Bloomington Streams over Time

This section of results relates to Research Question 1. Part (a), which is repeated below, addresses changes in Bloomington streams:

(a) Where and when have Bloomington streams been channelized and/or buried over time?

In total, the seven subwatersheds cover 58.93 square kilometers. Combined, the seven watersheds drain 97% of Bloomington’s land cover, with the remaining 3% draining Stephens Creek, Sinking Creek, and Leonard Springs (not included in this study). Figure 1a shows the major subwatersheds and the creeks and streams within Bloomington. All streams within Bloomington flow out of the city, with many stream beds being dry parts of the year. Five of these watersheds have the majority of their total area within Bloomington. Stout Creek and EFJC have the majority of their total area outside of Bloomington, but both watersheds contain several channels within Bloomington’s boundaries.

Figure 1b presents Bloomington’s hydrology in each watershed. The Hydrology layer shows both natural and artificial waterways. This includes streams, underground streams, canals, buried streams, and channelized streams. Most underground streams are in Stout Creek and the WFCC. Buried stream segments and channelized stream segments only occur in the Hydrology layer. These segments were found by comparing the Local Resolution Hydrology layer to the Creeks and Streams layer, the 2019 Bloomington Storm Outfall Map, and the GIS World imagery layer. Channelized stream segments and buried stream segments are shown in Figure 1b.

Figure 1b shows where Bloomington stream segments have been modified through stream burial or channelization. However, determining when stream channels have been modified requires inference based on when sections of Bloomington developed. Upper Clear Creek and Cascade developed first. Upper Griffy Creek, Upper Jackson Creek, and Stout Creek developed around the same time between 1939 and 1967. WFCC and EFJC developed after 1967 and before 2001.

Table 3. Changes in historical urbanization by drainage basin.

Basin	1939 Urban Area	1967 Urban Area
Griffy Creek	no	yes
WFCC	no	no
Stout Creek	no	yes
Jackson Creek	no	yes
EFJC	no	no
Cascade Creek	yes	yes
Clear Creek	yes	yes

shows the changes in Bloomington urbanization by watershed between 1939 and 1967. Watersheds that show evidence of urbanization in any part of the watershed are considered urbanized. Watersheds that do not have any urban development are considered not urbanized. Watersheds that had development in 1939, such as Cascade Creek and Clear Creek, experienced loss and fragmentation of historical UFPs, resulting in historically ephemeral patches in Upper Cascade Creek and Upper Clear Creek [5,42,44,46].

Table 4. Characteristics of stream channels and forest cover in the Bloomington drainage basins.

Basin	Percent Buried Streams (PBS)	Percent Channelized Streams (PCS)
Griffy Creek	10%	3%
WFCC	20%	26%
Stout Creek	17%	34%
Jackson Creek	28%	18%
EFJC	23%	20%
Cascade Creek	29%	22%
Clear Creek	52%	20%

shows each watershed’s percent buried streams and percent channelized streams. Attribute tables were compiled for each watershed’s hydrology lengths, channelized stream lengths, and buried stream lengths. Total segment lengths were calculated, with calculations shown in Table S1. Percentages of buried and channelized streams were calculated by dividing these total lengths by total hydrology lengths. Overall, most watersheds have a higher percentage of buried streams than channelized streams. Clear Creek has the highest percentage of buried stream segments because most of the urban core is in this watershed. Stout Creek and WFCC have a higher percentage of channelized streams, and Stout Creek has the highest percentage of channelized stream segments. This is due to the presence of an interstate highway that was constructed in these watersheds. Griffy Creek contains the lowest percentage of buried streams and channelized streams.

3.2. Stream Channel Connectivity to UFPs

This section of results relates to Research Question 1 (b), repeated below, which investigates stream impacts on stream connectivity to UFPs from channelization and burial in the city of Bloomington.

(b) How have these modifications altered channel connectivity to nearby UFPs?

Results from the i-Tree software’s selection of random points around Bloomington found that 271 out of 1000 points were identified as canopy cover. This is 27.1% of Bloomington’s area and covers about 16.4 square kilometers.

Table 5. Hydrological Tree Benefit Estimates found from estimated canopy cover in all watersheds of Bloomington, IN [46].

Benefit	Amount Kiloliter	±SE
Avoided Runoff	127.01	±6.59
Evaporation	2852.13	±147.93
Interception	2870.69	±148.89
Transpiration	2700.73	±140.07

shows the amount in kiloliters for the hydrological benefits these forests provide. Types of these benefits include avoided runoff, evaporation, interception, and transpiration. Avoided runoff benefits are calculated to be 7.748 kL/km²/year, and evaporation benefits are computed at 173.996 kL/km²/year. Interception benefits are estimated to be 175.129 kL/km²/year, and transpiration benefits are calculated to be 164.760 kL/km²/year. Only avoided runoff has a known method of economic evaluation. Avoided runoff refers to the prevention of water runoff, which reduces flooding. Evaporation and transpiration refer to water leaving the forest via water vapor. Transpiration only quantifies water exhaled by vegetation, whereas evaporation includes any type of water that leaves the forest as water vapor. Interception is the amount of water that is held in trees [45].

Figure 2 shows UFPs in Bloomington. UFPs shown on this map include areas that are 4047 m² (1 acre) or more, found using contiguous 30 m grid cells from NLCD forested land cover imagery that must contain at least 20% canopy cover [41,43]. Figure 2a shows UFPs along the Creeks and Streams layer, which shows that many of Bloomington’s UFPs are found near or along surface stream corridors.

Bloomington’s urban spatial matrix has been modified in such a way that stream channel connectivity to UFPs increases with distance from the uppermost reaches of five major watersheds in the community. These watersheds are Griffy Creek, WFCC, Jackson Creek, Cascade Creek, and Clear Creek. This is due to both the increasing amount of UFPs in downstream sections of these watersheds and the decreasing amount of buried or channelized segments as the channels flow out of the city [3,5]. Few UFPs are close to buried streams, as the stream gaps coincide with gaps in canopy cover. Figure 2b includes UFPs along with the Local Resolution Hydrology layer and shows fewer UFPs along waterways compared to the streams shown in Figure 2a.

The visual inspection of Bloomington UFPs showed that tree assemblages differed based on the UFP’s proximity to a stream and its geomorphology. UFPs along streams contained lowland assemblages. UFPs in valleys also tended to contain lowland assemblages, even if streams were not present. Lowland species include silver maple, sycamore, cottonwood, and swamp white oak. UFPs uphill from streams contained upland assemblages. Upland species include yellow poplar, sugar maple, hickory, and oak. Both types of patches include several species, which include red maple, black walnut, black cherry, and elm. Larger UFPs, like Griffy Woods, contain gradients of highland to lowland assemblages. UFPs in Stout Creek and WFCC tended to contain higher species richness, as the karst topography provides both lowland and highland habitat in proximity to each other.

Table 6. Characteristics of forested stream channels in the Bloomington drainage basins.

Basin	Percent Forested Hydrology (PFH)	Percent Forested Stream (PFS)
Griffy Creek	66%	82%
WFCC	32%	61%
Stout Creek	42%	56%
Jackson Creek	33%	54%
EFJC	30%	50%
Cascade Creek	27%	49%
Clear Creek	12%	29%

shows each watershed’s percent forested stream length and percent forested hydrology length. Attribute tables were compiled for each watershed’s hydrology lengths, forested hydrology lengths, forested stream lengths, and creek and stream lengths. Total segment lengths were calculated, with calculations shown in Table S1. Percent forested hydrology was calculated by dividing these total lengths by total hydrology lengths. The percentage of forested streams was calculated by dividing total forested stream lengths by total creek and stream lengths.

The percentage of forested streams is always higher than the percentage of forested hydrology. This is due to the stream layer containing only surface streams, as shown in Figure 2a. Many surface stream corridors are surrounded by UFPs due to being high-risk areas for flooding. The forested hydrology map shown in Figure 2b includes buried streams and channelized streams but omits most channels less than 12 m wide. Buried and channelized stream corridors are often surrounded by ISC and receive rainfall that contains little to no exposure to UFPs before flowing into the stream corridor. These differences account for most of the lower percentages seen in forested hydrology. However, the category of forested streams includes lakeshores as stream surface, which overestimates the percentage of forested streams in Griffy Creek. Overall, Griffy Creek has the highest percentage of forested streams and hydrology. Clear Creek contains the lowest percentage of forested area in both categories.

3.3. Urban Forest Patch Change over Time

The results below relate to Research Question 2 (a), repeated below, which addresses changes in UFPs in Bloomington between 1939 and the present.

(a) Where and when have Bloomington UFPs changed over time?

Figure 3 shows Bloomington aerial imagery from 1939 and 1967, respectively. Forested areas can be seen as darker sections within the images. Most areas surrounding the Bloomington borders in 1939 were agricultural fields and had been deforested to clear land and provide timber for a large local furniture company before these areas experienced urbanization [46]. In the 1967 imagery, some reforestation can be seen in former (1939) agricultural areas, especially along riparian zones. Urbanization has mixed results for streams and forests. Initial developments fragmented most of the remaining forests, which resulted in remnant patches. Eventually, many of these patches stabilized into environmental resource patches. Additionally, the city and urban landowners provided spaces that grew new UFPs or expanded existing UFPs. These became a mixture of additional environmental resource patches and introduced patches based on species composition and their locations in the Bloomington landscape [5]. Some areas, such as Griffy Creek and WFCC, increased in forested area after urban development due to formal and informal protection [47].

Figure 4 shows Bloomington ISC in 2001 and 2019. Slight changes in ISC occurred between 2001 and 2019, with noticeable increases in ISC being observed in parts of the WFCC and EFJC. The NLCD ISC classifications are used to categorize groups by percent ISC. Open space areas (OSAs) contain less than 20% ISC and are shaded black or blue. Low-intensity developed areas (LIDAs) are defined as having a percent ISC between 20% and 49% and are shaded purple or pink. Medium-intensity developed areas (MIDAs) have a percent ISC between 50% and 79% and are shaded orange or yellow. High-intensity developed areas (HIDAs) have 80% or greater ISC and are shaded yellow [48]. Areas shown in black in Figure 4b, which symbolize forests, waterbodies, and open fields, often match up with UFPs shown in Figure 2. Comparisons between Figure 4a,b show little change in the Bloomington landscape matrix between 2001 and 2019, based on minimal differences in ISC [5,44,49].

3.4. Stream and Urban Forest Patch Connectivity over Time

The results in this section relate to Research Question 2 (b), repeated below, which addresses how changes in the UFP area have impacted streams in Bloomington between 1939 and the present.

(b) How have these changes in the UFP area altered their impact on streams?

Table 7. Characteristics of forest cover and ISC in the Bloomington drainage basins.

Basin	Percent UFP Cover	2001 Average Percent ISC	2019 Average Percent ISC
Griffy Creek	59%	7%	9%
WFCC	33%	18%	25%
Stout Creek	26%	20%	24%
Jackson Creek	17%	23%	27%
EFJC	12%	22%	27%
Cascade Creek	19%	26%	29%
Clear Creek	9%	37%	40%

shows each watershed’s percentage forest cover, the 2001 average percent ISC, and the 2019 average percent ISC. To calculate the percentage of forest cover, attribute tables were compiled for each watershed’s UFP area and watershed area. These tables are shown in Table S1. Furthermore, the UFP areas were divided into their respective watershed areas. Average percent ISCs were obtained from each watershed’s Arc GIS Pro raster metadata, shown in Table S1.

Overall, Table 7 shows that watersheds with a higher percent UFP have a lower average percent ISC. Griffy Creek Watershed contained the highest percent UFP and lowest ISC in both 2001 and 2019. Clear Creek Watershed contained the lowest percent UFP and the highest percent ISC in both 2001 and 2019. The average percent ISC in 2001 ranged from 7% to 37%. All watersheds increased between 2001 and 2019 in average percent ISC, with increases varying from 2% to 7%, but all watersheds during this time period ranged between 9% and 40% average percent ISC. Therefore, all watersheds have average classifications of OSA or LIDA [49]. These are lower ISC averages than in other cities with ISC studies [48,50]. However, all watersheds have a high enough average ISC within their landscape matrices to suggest stream ecosystem degradation [5,51,52]. Watersheds that have more than 5% average ISC show a decline in stream ecosystem health. All Bloomington watersheds have an average ISC greater than 5%. Once the average ISC reaches 15%, the watershed could lose approximately 60% of its benthic invertebrate fauna. Only Griffy Creek had less than 15% average ISC in both 2001 and 2019, which suggests that the other six watersheds contain heavily degraded stream ecosystems. Griffy Creek contained 7% average ISC in 2001 and 9% average ISC in 2019, which suggests that this stream’s ecosystem is in decline but is more intact than any other Bloomington watershed [51,52].

The ranking of Bloomington’s seven main watersheds helps quantify overall stream-to-UFP connectivity for each Bloomington watershed, which addresses both questions 1 (b) and 2 (b). The watershed ranking is based on the information provided in Table 3, Table 4, Table 6 and Table 7, with the ranking values shown in

Table 8. Ranking of drainage basins to show connections between stream and forest characteristics based on Table 3, Table 4, Table 6 and Table 7.

Basin	1939 Urban Area	1967 Urban Area	PBS	PCS	PFH	PFS	Percent UFP Cover	2001 Average Percent ISC	2019 Average Percent ISC	Score
Griffy Creek	0	1	1	1	1	1	1	1	1	8
WFCC	0	0	3	5	4	2	2	2	3	21
Stout Creek	0	1	2	6	2	3	3	3	2	22
Jackson Creek	0	1	5	2	3	4	5	5	4	29
EFJC	0	0	4	3	5	5	6	4	4	31
Cascade Creek	1	1	6	4	6	6	4	6	5	39
Clear Creek	1	1	7	3	7	7	7	7	6	46

. This table assigns a value to the rank order among study streams for each characteristic listed below. Watersheds are ranked from lowest score to highest score, with scores calculated in Table S1. Watersheds receive a rank from 1 to 7 for buried stream segments, channelized stream segments, percent forested hydrology, percent forested stream, percent UFP cover, 2001 average percent ISC, and 2019 average percent ISC. Historical urbanization receives a 1 for presence or a 0 for absence. Griffy Creek received the best score overall. Griffy Creek has been partially urbanized since 1967, but most of the watershed is undeveloped forest. WFCC ranked second, with a high percent of UFP cover, a high percent of forested streams, later development, and a low percent of buried streams. Stout Creek ranked third due to having a high percentage of channelized streams, a low percentage of buried streams, and a moderate amount of UFP cover. Jackson Creek ranked fourth, mostly due to a high percentage of buried streams but with a high proportion of forested hydrology and a low percentage of channelized streams. EFJC ranked fifth due to a low percent of UFP cover, a low percent of forested hydrology, a low percent of forested streams, and later development. Cascade Creek ranked sixth due to its low percent of forested streams, low percent of forested hydrology, and high percent of buried and channelized streams. Finally, Clear Creek ranked last. It has the most buried streams, the lowest percent UFP cover, the lowest percent forested streams, the lowest percent forested hydrology, a high average percent ISC, and was the only watershed to be mostly urbanized in 1939.

Table 9. Watershed rankings based on results from Table 3, Table 4 and Table 5.

Watershed	Total Score	Rank
Griffy Creek	8	1
WFCC	21	2
Stout Creek	22	3
Jackson Creek	29	4
EFJC	31	5
Cascade Creek	39	6
Clear Creek	46	7

shows Bloomington watersheds ranked in order from most to least connected and each watershed’s total score.

4. Discussion

4.1. Stream and Urban Forest Patch Locations

Past and present locations of streams and UFPs address Questions 1 (a) and 2 (a), respectively. The locations of current streams are typically determined by previous channel modifications [2,3,5,22,23]. The directional flow of Bloomington streams makes both the community and UFPs vulnerable to potential periods of drought and less water-secure than downstream communities. Many streams in the upper sections of Bloomington are dry stream beds for part or most of the year. Historically, the lack of consistent streams caused water insecurity in Bloomington in the late 1800s and early 1900s. Water was only pumped into the main lines once or twice a week. Additionally, water would be brought in via the railroad. In order to provide a more consistent water supply, Indiana University and Bloomington created University Lake in 1914 by damming a branch of Griffy Creek. Griffy Lake was later developed in 1924 by damming the main branch of Griffy Creek downstream of University Lake. Currently, Bloomington sources water from Lake Monroe in a neighboring community, but the construction of wetland UFPs could increase Bloomington’s water security [3,19,33,46,53,54].

Past and present locations of UFPs are based on general land use, ownership, and governance of Bloomington’s landscape [3,5,47]. Historically, all of Bloomington’s landscape supported dense oak/hickory forests [46]. Currently, almost all of Bloomington’s UFPs are found outside of the urban core. This means that most of the UFP flood reduction and carbon sequestration occurs in patches in the suburbs and outskirts of the city. This does not remedy potential flooding or pollution in or around the urban core. Since all Bloomington streams flow away from the city and most buried streams are found in the urban core and major shopping centers, downtown Bloomington and the major Bloomington shopping centers of Whitehall Mall and College Mall are the sites most likely to flood after moderate to heavy precipitation. Two such incidents occurred in 2008 and 2021, with each incident causing major property damage and power outages, as well as putting human lives at risk due to rapidly moving water, exposure to water-borne pathogens, and the risk of drowning. Flooding from downtown and major shopping centers moves downstream to areas of Clear Creek, Cascade Creek, Jackson Creek, and WFCC. Jackson Creek and WFCC have numerous UFPs that may reduce flooding in these downstream areas compared to Clear Creek and Cascade Creek. While Cascade Creek has Cascade Park, this UFP is separated from Cascade Creek with ISC in the form of trails, roads, infrastructure, a parking lot, and a rock-armored stream. Clear Creek has few UFPs to reduce downstream flooding and would benefit the most from the implementation of green spaces, green infrastructure, and UFPs, especially considering the flooding events of 2008 and 2021 [5,9,19,20,31,32,45,55].

4.2. Importance of Connectivity

Investigating urban stream-to-UFP connectivity builds upon Forman’s landscape ecology theory of patches by looking at the socioecological relationship between two different types of patches in a landscape matrix. In particular, it looks at how the chronic disturbance of urban development affects stream corridors and UFPs and their interactions [3,5]. Urban development and increasing ISC in the landscape matrix act as chronic anthropogenic disturbances that alter both ecosystems and interactions for urban streams and UFPs [2,3,5]. For stream corridors, the disturbance of urban development and increasing ISC cause decreases in species diversity and overall water quality [2,4,51,52]. For UFPs, the disturbance of urban development leads to increased amounts of invasive species, a decreased ability to sequester atmospheric nitrogen, and a decreased diversity of soil microbes [11,12].

Eventually, stream corridors and UFPs reach a new equilibrium resulting from changes in connectivity between each urban stream corridor and their respective UFPs [3,5]. This connectivity is heavily dependent on the amount of development on or near the land that these ecosystems occupy. For streams, the amount of channelization and burial are the largest determining factors for connectivity with riparian forest cover [3,5,19,23,56]. Streams with high connectivity have fewer anthropogenic alterations in their watersheds than streams with lower connectivity. This relates to both in-stream modifications addressed in Question 1 (b) and watershed modifications addressed in Question 2 (b) [2,4,18,19].

In-stream modifications include both channelization and stream burial. Channelization decreases drainage time for watersheds. However, this also reduces the ability of water to interact with the surrounding environment. Buried streams have the least connectivity with UFPs because these streams are mostly unable to interact with their surrounding environment. Most of these streams were seen in Clear Creek, Cascade Creek, and Jackson Creek. Channelized streams include streams that flow through concrete-lined banks and streams that have their sides covered in rip-rap. Clear Creek and Cascade Creek contain the more concrete-lined banks, but the highest proportions of channelized streams were Stout Creek and WFCC, with most of these streams occurring as rip-rap streams [2,4,18,57,58]. The reduction of stream to UFP connectivity contributes to the symptoms of urban stream syndrome, which include altered channel morphologies, increased nutrient and chemical loading, reduced species richness, and an increase in the number of tolerant species. Anthropogenic modification of streams may contribute to flooding directly downstream during most rain events and may back up to cause flooding in the urban core [2,4,57,58]. The streams with higher connectivity are typically in less developed areas of suburbia. Many of these stream reaches contain larger UFPs surrounding the stream and have less channelization. These watersheds, including Griffy and Lower Jackson Creek, usually contain publicly protected UFPs. Other highly connected streams flow through private UFPs with areas of karst topography, such as those in Stout Creek and lower WFCC. These watersheds are protected by difficult-to-develop areas and the presence of environmentally sensitive habitat. Additionally, stream reaches with perceived aesthetic or recreational value are usually less channelized [17,23,30,59].

Watershed modifications include changes in UFPs and ISCs in the urban landscape matrix [5]. UFP connectivity to streams is dependent on several factors. The proximity of UFPs to the stream contributes the most to connectivity. Connectivity can be impaired if ISC separates the UFP from the stream. UFP’s connectivity to a stream can be reduced if the forest soil is covered by ISC. This can occur through the placement of paved trails, bridges, or other types of ISC. ISC increases storm runoff within UFPs and reduces the ability of UFPs to mitigate storm runoff. Additional factors that reduce UFP connectivity include lower canopy cover and UFP fragmentation. Having fewer trees that are not in a connected forest patch reduces the amount of water that can be intercepted and provides less evaporation area. UFP fragmentation also reduces the amount of forested area that connects with streams [5,9,19,45,58]. Additionally, buried streams lower local water tables, which reduces UFP connectivity with streams, especially for upland UFPs. These ecosystems often become drier and can inhibit tree growth due to the lack of root access to water. Buried streams also reduce infiltration in these patches. These lowered water tables force riparian soils to become aerobic, which reduces denitrification in riparian UFPs. This causes some riparian UFP to become sources for nitrogen instead of sinks [3,59].

Separating streams and UFPs negatively affects both the stream and the UFPs. Disconnected streams and UFPs suffer from similar forms of ecosystem degradation. This includes increased contaminant loading, decreased species richness, habitat fragmentation, lowered water tables, and possible ecosystem collapse [2,9]. However, other symptoms of separated streams and UFPs differ. For separated streams, both the physical and chemical characteristics are changed. In particular, these streams have faster rates of flow and increased amounts of nutrients [23]. For separated UFPs, the changes include hydrologic drought, which manifests as limited tree growth in upland UFPs due to reduced soil moisture. In riparian UFPs, hydrologic drought reduces species richness and denitrification. These forests can become sources of nitrification as wetland obligate species decay [3,55]. Overall, connecting streams to UFPs is important because streams rely on UFPs to reduce stormwater runoff, provide habitat and allochthonous inputs, and reduce contaminant and nutrient loading. Urban streams with few UFPs are more prone to flooding and severe habitat degradation [4,15,18,56]. Connecting streams to UFPs is especially important for upland UFPs, which may remain unconnected to lowered water tables or UFPs in more drought-prone areas. Riparian UFPs will still have access to the water table, but they may acquire the characteristics of upland forests [3,60,61].

Urbanization does not always cause degradation in connectivity between streams and forests. If rural areas have been previously cleared for natural resource extraction or agriculture, urban development can maintain or increase connectivity between streams and forest patches in a landscape. This depends on the changes in land use around environmental resource patches. New UFPs have emerged due to the urbanization of former fields, especially in flood plains, which increases stream-to-UFP connectivity. Before Bloomington expanded into the Jackson Creek watershed, the watershed was used for agriculture. This usage has removed most forests to grow as many crops as possible. After annexation, these lands saw forests emerge in spaces between infrastructure and in new designated spaces. A few current UFPs remain intact from before annexation. One of these forests is Latimer Woods, which has remained almost unchanged since 1939. Crucial factors include designated land usage for UFPs and streams and governance of these resources [5,24,46,61,62].

4.3. Improving Connectivity between UFPs and Streams

To improve the allocation of Bloomington UFP ecosystem services, UFPs need to be protected, or newly planted, in areas where streams are flowing out of a highly channelized section of their reach. Additionally, existing UFPs can be managed for continued resilience and persistence using urban silvicultural techniques [63]. This allows UFPs to continue to provide ecosystem services while also benefiting urban streams, which then provide ecosystem services to the UFPs. Bloomington is an atypical city, as no streams flow into the city. This makes the better allocation of UFP services limited to areas directly outside of the urban core. Most of these areas currently lack sufficiently sized UFPs to reduce the symptoms of urban stream syndrome and often have highly channelized surface streams. Streams must be directly remediated to regulate flooding [2,4,19,57,58,60]. Areas vulnerable to flooding are directly north and south of the urban core and south of College Mall and Whitehall Mall. Only one of these sites has a large enough UFP to provide hydrological ecosystem services to abate flooding. This is Latimer Woods, south of College Mall. It currently reduces runoff for a high-ISC drainage area in Jackson Creek. Other methods to reduce runoff are to break up sites with large uninterrupted SCs with added green infrastructure and to reduce the area where buried streams drain. Most areas with high, uninterrupted ISC coincide with artificially buried streams. Breaking up these parking lots with greenspaces or pervious landcover, daylighting streams, and planting street trees could help alleviate potential flooding [4,19,64].

5. Conclusions

Streams and forests are connected through the movement of water, but urban areas often reduce or disrupt this connectivity. Urban streams are often channelized, and UFPs can be isolated by both their distance from streams and interruption by ISC [2,9,65]. Here, seven watersheds in Bloomington were ranked using hydrology, tree canopy, UFP, ISC, and historical aerial imagery. Watersheds with greater forested stream length were found in areas with lower percentages of ISC, in areas with few buried streams, and in areas that have been developed more recently. Conversely, areas with the highest percentages of ISC contained the most buried streams and highly channelized surface streams and were found in historically developed areas in the urban core. Stream burial and channelization occurred concurrently with the development of an area. These sites also contain little tree canopy and few UFPs. Many UFPs along streams have emerged more recently, as most of Bloomington’s annexations have incorporated mostly former agricultural land with few forest parcels. Forest parcels that were present in 1939 often became fragmented UFPs by 1967. As Bloomington continues to develop and expand into currently forested surrounding areas, many unprotected UFPs will disappear unless there are efforts to protect them. This may take the form of public ownership with conversion into parks or green infrastructure, purchase by land trusts, private ownership and protections, or combinations of public and private ownership protections. These forested areas include riparian UFPs, which are often owned as a patchwork of public and private parcels. These UFPs could degrade and eventually disappear if public or private entities decide to channelize a stream for urban development [3,4,23,24,57].

Few studies have examined relationships between UFPs and urban stream ecosystems. Studies observing forest and stream interactions either take place in rural areas, analyze habitat connectivity, or focus on an urban forest system’s ability to provide green stormwater infrastructure [3,14,17,19,20,62,64]. Future research could look at the relationship between streams and UFPs in larger cities. Many of these cities contain multiple urban cores, commercial areas, larger streams, and inflowing streams [3]. Unlike Bloomington, UFPs in suburban regions upstream of an urban core would likely benefit the urban core. Other areas of future research should include the combined ecosystem services of forested streams beyond flood regulation, economic evaluations of these ecosystem services, and the relationships between UFPs and urban streams in varying ecoregions [6,18,66].