A Habitat-Based Framework for Communicating Natural Resource Condition

Progress in achieving desired environmental outcomes needs to be rigorously measured and reported for effective environmental management. Two major challenges in achieving this are, firstly, how to synthesize monitoring data in a meaningful way at appropriate temporal and spatial scales and, secondly, how to present results in a framework that allows for effective communication to resource managers and scientists as well as a broader general audience. This paper presents a habitat framework, developed to assess the natural resource condition of the urban Rock Creek Park (Washington, DC, USA), providing insight on how to improve future assessments. Vegetation and stream GIS layers were used to classify three dominant habitat types, Forest, Wetland, and Artificial-terrestrial. Within Rock Creek Park, Forest habitats were assessed as being in good condition (67% threshold attainment of desired condition), Wetland habitats to be in fair condition (49% attainment), and Artificial-terrestrial habitats to be in degraded condition (26% attainment), resulting in an assessed fair/good condition (60% attainment; weighted by habitat area) for all natural resources in Rock Creek Park. This approach has potential to provide assessment of resource condition for diverse ecosystems and provides a basis for addressing management questions across multiple spatial scales.


Introduction
One of the key challenges of large-scale monitoring programs is to develop integrated and synthetic data products that can translate a multitude of diverse data into a format that can be readily communicated to decision-makers, policy developers, and the public [1][2][3]. Such timely syntheses of ecosystem condition can provide feedback to managers and stakeholders, so that the effectiveness of management actions as well as future management goals can be determined at multiple scales [4].

Integrative Indices and Report
Cards. One approach to synthesizing data has been the development of multimetric indices to summarize the status of a community, such as stream fish, and then draw inferences on the status of the supporting ecosystem [5]. Metrics such as the fish index of biotic integrity (FIBI) and the benthic IBI have been widely applied, both internationally and regionally (e.g., streams in Maryland, USA). IBI metrics are seen as providing greater insight into ecosystem condition than physical measurements (e.g., water quality) alone, as biological communities provide an integrated summary of ecosystem condition over time [6][7][8]. However, in the absence of a rigorous process for integrating data on diverse biotic communities and other ecosystem components and for communicating results to decision-makers, these indices and bioindicators have had limited effectiveness in improving ecosystem management [9].

ISRN Ecology
Environmental score cards or report cards are seen as an important tool for this type of integrated assessment, to move beyond simply identifying ecosystem change and on to applying monitoring data to ecosystem management [10,11]. In some cases, assessments of multiple metrics using threshold-based environmental report cards have been used to compare estuaries at broad spatial scales [12][13][14][15]. At the local level, however, most environmental assessments still tend to focus on a few, specific resources rather than providing an integrated overview of site conditions.

Consideration of Scale in Integrated
Assessments. An important consideration in determining the appropriate spatial and temporal scale of an integrated assessment is identifying the characteristic scales of variability in the ecological resources, as well as in the potential stressors to those resources.
Environmental processes typically show a direct relationship between temporal and spatial scales, with fine-scale spatial processes happening rapidly and broad-scale processes happening slowly [16,17]. Many stressors also follow this pattern, with local impacts such as point-source inputs being short term and concentrated in small areas while global impacts such as climate change occur slowly over large spatial scales [18] (Figure 1). Less intuitive, and more difficult to measure, are shifting baselines such as nutrient increases, where gradual changes occur at even small spatial scales [19] and extreme events such as hurricanes where stressors can impact large areas in a very short period of time ( Figure 1). In some cases, the aggregation of multiple small-scale stressors can also produce larger-scale impacts ( Figure 1). Time versus space plots (Stommel diagrams) have been used extensively [17,20] to illustrate these scaling relationships for different organisms and processes in oceanic [21] and terrestrial [16] environments.

Monitoring Data Collected at Multiple Scales.
Integrative monitoring programs typically collect data on different variables at a variety of spatial and temporal scales. The spatial scale of measurement is important to consider for effectively combining metrics into integrated assessments of resource condition. A large number of metrics can be measured over small spatial scales for relatively little expense (e.g., biodiversity and water quality), so it is often useful to aggregate these metrics to get a more stable measure of condition ( Figure 2). Aggregation can be accomplished by the mathematical combination of multiple metrics, for example, into a water quality index [22,23] or the benthic index of biotic integrity (BIBI) [24]. Metrics providing information at very large spatial scales are often considered individually within an integrated assessment because they are typically more difficult or costly to obtain or represent one number for a large area such as percent impervious surface [25] (Figure 2).
The temporal scale of metric measurement is also important to consider for data synthesis. Metrics that change over short temporal scales can be measured at high frequency, although interpreting the high data variance can be challenging; in one example, rolling averages have been developed for ozone [26]. For metrics related to long-term trends, measures of central tendency are often suitable; longterm monitoring of land cover change often relies on key spatial pattern indices such as mean patch size, for example [27]. Matching the scale of observation to the scale of environmental change remains a central challenge in ecology [28]. Not only careful metric selection but also a strong framework to integrate diverse metrics, collected at different spatial and temporal scales, can assist in the interpretation of trends in natural resource condition and explanation of linkages between condition and diverse stressors.

Aim.
The aim of this study was to integrate a series of available monitoring metrics to assess the current resource condition of habitats within Rock Creek Park and to identify ways to improve future assessments of resource condition.

Study Site. Rock
Creek Park is located on approximately 710 hectares in the District of Columbia, USA, at the downstream end of a highly urban and continually developing Piedmont river valley [29]. The Park was established in 1890 at a significant time in the development of what would later become the National Park System. As a result, the protective prescription for the Park called for regulations to "provide for the preservation from injury or spoliation (plundering) of all timber, animals, or curiosities within said park, and their retention in their natural condition, as nearly as possible" [30,31]. Therefore, determination of key habitats and assessment of resource condition is particularly  relevant to the management of Rock Creek Park, a unit within the National Park Service.

Determination of Assessment
Scale. The first step in the assessment process was to identify the spatial and temporal scale for assessment. Relevant scales for integrated assessments of natural resource condition may vary from 100s of meters, for example, within a managed land such as a national park or between subwatersheds, to 1000s of kilometers, for regional or even global-scale comparisons. Different data and approaches have been applied at these different scales [22,32,33]. In consultation with Park management staff, the assessment scales were determined based on potential management questions and interpretation as well as data availability. The spatial extent for the assessment was established as the current Park boundary (classified as the Fee Boundary) and the temporal extent is an eight-year period from 2000 to 2008.

Identification of Habitats.
Many ecological classification systems exist, based on such features as vegetation communities [34,35] or land cover [36]. In the current assessment, a classification defined by the International Union for the Conservation of Nature (IUCN) was used, to facilitate potential comparisons to other ecological systems [37]. The IUCN habitat classification includes 16 habitat types at the highest level, including both terrestrial and aquatic habitats, desert, forest, and subterranean cave habitats.
To determine the habitat types present within Rock Creek Park, a classified vegetation base map, from the National Capital Region (NCR) Inventory and Monitoring Program (I&M), was aggregated into general habitat types: beach/mixed oak, beach/tulip poplar, beach/white oak, chestnut oak, loblolly pine/mixed oak, tulip poplar, Virginia pine/oak, and sycamore/green ash were classified as "Forest habitat"; streams, seeps, and springs were classified as "Wetland (inland) habitat"; and canopy gap, meadow, mowed lawn, and mowed lawn with trees and shrubs were classified as "Artificial-terrestrial habitat". Although discussions with resource managers originally determined a longer list of habitats (e.g., multiple forest types), available data density did not allow assessment at this level of precision, so the higher level habitat classification (three habitats) was used for condition assessment calculations.

Selection of Metrics.
Within the habitat assessment framework, metrics were chosen to establish a definition of the conceptual range of habitat condition, from desired to degraded, which were applicable to management goals and objectives (     the National Atmospheric Deposition Program (Table 1). All metrics are used within the NPS I&M program and hence have been through a stringent process of protocol development as well as careful assessment for responsiveness to ecosystem changes [3]. Seven metrics were used to assess each habitat and two metrics (deer population and percentage of impervious surface) were used in the assessment of both Forest and Artificial-terrestrial habitats, resulting in a total of 19 separate data sets. Data used for the assessment of Rock Creek Park were collected between 2000 and 2008, with data density ranging from one park-level value at one time (e.g., impervious surface, critical connectivity) to 241 total measurements (e.g., dissolved oxygen, monthly measures from 13 sites) ( Table 2).
Five of the indicators used to assess ecosystem condition of Forest habitats were measures of biodiversity, with two indicators of ecosystem processes (Figures 2 and 3). Of the seven indicators, three are relevant to the measurement of 6 ISRN Ecology  [46,47] see also [48] Presence of forest interior dwelling species (FIDS) of birds >1 "highly sensitive" FIDS >4 "sensitive" FIDS [49] see also [50] Deer density Forest: < 8 deer km −2 [51] see also [52,53] [74,75] also see [72,76] Total NO 3 deposition (cultivated) <10 kg ha −1 yr −1 [75,77] also see [72] Hg deposition <2 ng L −1 [78][79][80] the quality of a resource within Forest habitats (presence of forest interior dwelling birds, native seedling regeneration, connectivity), while the remaining four (exotic tree/shrub density, presence of pest species, deer density, impervious surface) measure stressors that are likely to directly impact Forest habitats. Six of the seven indicators used to assess condition of Wetlands habitats were water quality indicators: three indicators of stressors (total phosphorus, salinity, aqueous nitrate) and three indicators of habitat quality for aquatic fauna (dissolved oxygen, benthic IBI, physical habitat index). The remaining indicator (proportion of area occupied by adult amphibians) is a measure of biodiversity and provides information regarding the ecosystem resource value of the habitat. Assessment of Artificial-terrestrial habitat condition was based on indicators in three categories: biodiversity (deer density), ecosystem processes (impervious surface), and air quality (soot, ozone, sulfate deposition, nitrate deposition, mercury deposition) ( Figure 2, Table 2).
Recognizing that air quality indicators could be used for any habitat, particularly any terrestrial habitat, they were considered particularly appropriate for use in the predominantly mowed grass characterizing Artificial-terrestrial habitat within Rock Creek Park ( Figure 2, Table 2). These open vistas rely most heavily on the resource of high visibility and are likely to be more directly impacted in terms of plant health by high ozone, due to plants being more exposed in open grassland than a closed forest. Wet deposition of sulfate, nitrate, and mercury is more likely to have a negative impact in open areas where transport rate to groundwater and creeks will be greater than areas such as forest where more rainfall is intercepted by plant canopies and where there is greater plant biomass to reabsorb nutrients from soils.

Definition of Thresholds and Calculation of Condition
Assessment. To assess the natural resource condition of Rock Creek Park, thresholds were established for the 19 separate metrics using scientific literature, management goals, regulations, and professional judgment where necessary ( Table 2).
In nearly all cases these thresholds were justified ecologically by the scientific literature, even though many of them were set as either management or regulatory thresholds ( Table 2). For the seven metrics identified to assess each habitat, data were compared to these threshold values. The percentage of sites and times attaining the threshold value for each metric was calculated, where a value of 100 indicated that all sampling sites and times met the threshold to maintain natural resources (desired condition), and a value of zero indicated that no sites at any sampling time met the threshold value (degraded). The attainment of threshold condition for each of the three habitat types present within Rock Creek Park was calculated as an unweighted mean of the attainment scores for the seven metrics used to assess the condition of that particular habitat. Calculation of the park condition status was calculated as an area-weighted mean, based upon the relative area of each habitat type within Rock Creek Park. For determination of status of metrics, habitats, and the whole park assessment, percentage attainment scores were categorized on a uniform scale from very good (81-100% attainment) to very degraded (0-20% attainment).

Results
Three habitats were defined within Rock Creek Park; Forest habitat making up 81% of the total area (575 ha), Wetland habitat 2% (14 ha), and Artificial-terrestrial habitat comprising the remaining 17% of Park area (121 ha) (Figure 4). Within the habitat assessment framework, the threshold attainment scores for individual metrics ranged from 0% attainment (very degraded) to 100% attainment (very good) ( Table 3). The overall, area-weighted, habitat-based condition of Rock Creek Park, based on attainment of ecological threshold for 19 metrics, was assessed as fair to good (60% attainment; Table 3, Figure 5).

Forest Habitats. Forest habitats within Rock
Creek Park were assessed to be in good condition (67% attainment of threshold condition; Table 3  (mean ± standard error; 0% attainment) and low native seedling regeneration of 6, 000 ± 3, 221 seedlings ha −1 (0% attainment). However, the Park had generally low but highly variable exotic tree and shrub density (16.7 ± 10.9% of total basal area; 70% attainment), low impervious surface (5% of land surface impervious), and high forest patch connectivity, with the Park achieving critical connectivity (D crit ), by assuming that patches within 340 m of one another are connected (100% attainment within the park). The Park also had low forest pest species (none observed) and diverse forest interior dwelling bird species (100% attainment).

Discussion
Effective ecosystem management relies on clear, synthetic communication of natural resource condition to managers, decision makers, and the interested public. The habitat  framework developed for the assessment of natural resource condition in Rock Creek Park provides a framework for the synthesis of diverse metrics to allow integrated assessments at multiple spatial and temporal scales, presenting results in a readily understandable and communicable format.

Implications of Rock Creek Park
Assessment. An overall fair/good assessment of natural resource condition for the entire Park (60% attainment of ecological threshold condition) reflected the large area of Forest habitat in good condition the relatively small area of Artificial-terrestrial habitat (mostly mowed grasslands) in degraded condition, and the very small area of wetland in fair condition (Table 3, Figure 5). The assessment of Forest habitat within the Park was good; however, this habitat is clearly challenged by very high deer populations and low native seedling regeneration (Table 3, Figure 5), suggesting that the Forest is susceptible to further degradation. Species richness and abundance of herbs and shrubs have shown measurable reductions at densities as low as 3.7 deer km −2 and are consistently reduced as densities approach 8.0 deer km −2 [81]. Densities of 10-17 deer km −2 inhibited the regeneration of understorey species in a large manipulation study, whereas a diverse understorey was supported at 3-6 deer km −2 [82]. It is therefore likely that high deer populations recorded for Rock Creek Park (approximately 26 deer km −2 ) are related to the low native seedling regeneration ( Table 3). Measurement of exotic tree and shrub density was highly variable (Table 3) with a mean value three times the threshold of 5% of total basal area, suggesting patchy but intense infestations. This metric was assessed from only three forest sampling plots (Table 2) and so increased spatial assessment would be valuable to more fully assess the potential impact of exotic plants on forest condition. Forest connectivity was close to the threshold (340 m measured, 360 m threshold; Table 3), suggesting that the Forest is susceptible to further fragmentation within the Park (by roads and trails, etc.), potentially reducing the habitat value for both flora and fauna.
High nutrient concentrations, depauperate benthic stream communities, and high salinity resulted in an assessment rating of fair for Wetland habitats within Rock Creek Park (Table 3). Nitrate concentration has been found to be negatively correlated to the integrity of benthic communities [24], and the statewide mean nitrate concentration of 2.45 mg L −1 is lower than the mean of 2.6 ± 0.1 mg L −1 measured within Rock Creek Park. Rock Creek watershed is also highly urbanized and the integrity of benthic communities has previously been correlated negatively to urbanization within the surrounding watershed [83]. Salinity in Rock Creek and tributaries was greater than 0.25 g L −1 in 70% of samples, with salinities reported as high as 1.7 g L −1 during winter, when road salting occurs [29]. Although short-term exposure (96 hrs) to salinities as high as 10 g L −1 has been found to have little effect on individuals of multiple macroinvertebrate species [84], many studies have reported biotic degradation or species loss and promotion of exotic species at salinities greater than 1.0-1.4 g L −1 [85,86]. There has been a regional (Maryland, USA) increase in salinity over the past 30 years [87], so salinity has high potential to pose an increasing threat to the water quality in Rock Creek, with its highly urbanized watershed.
The degraded assessment for Artificial-terrestrial habitats within the Park resulted from the desired condition being largely based on air quality (Figure 3), with all five metrics of air quality having very low threshold attainment (Table 3). This raises the issue of how to define desired condition, especially for altered habitats that still have ecosystem values. In this case, for example, the Artificial-terrestrial habitats are mostly mowed grasslands-but with best management they could contain meadow areas habitat for insects and buffer areas along streamlines to limit nutrient inputs. This indicates a data gap, as no metrics are currently monitored to address changes in native meadow to mowed grass or extent of these data, which would allow an improved assessment of the natural resource condition of Artificial-terrestrial habitats, atmospheric metrics, which indicate more regional challenges.

A Habitat Framework Can Assist in the Assessment
Process. Scientific understanding of ecosystem processes is often limited to the temporal and spatial scales of the research, requiring clear frameworks to facilitate translation of understanding across scales and application by decisionmakers and managers [88]. Assessments of ecosystem status are increasingly being recognized for their value to direct and assess management actions [3,4]. However, as the spatial scale of impacts becomes larger (e.g., air quality and climate change), the challenge is to develop frameworks that will allow comparison between different locations, or management units, as well as across spatial and temporal scales.
The condition and trend of natural resources can be evaluated over multiple temporal and spatial scales (Figure 1). Global [89] and national [90] level assessments have long been a focus of attention and regional level assessments have become increasingly prominent [23,91]. At the local level, however, most environmental assessments still tend to focus on a few, specific resources rather than providing an integrated overview of site conditions. The most common framework for conducting ecosystem assessments is a geographic basis, for example, estuaries along a coastline [13] or regions within a larger system [92]. This is ideal when the different regions can be measured with the same metrics and are essentially similar (e.g., all temperate forest, all estuarine habitats, etc.). However, when disparate comparisons are requested (e.g., tropical and temperate estuaries, coastal and mountain landscapes), different ecological drivers may be present and the various communities may interact differently. One approach to dealing with this variability is to assess on a species by species basis; careful choice of indicator species can provide valid assessments of some aspects of an ecosystem [93]. However, an indicator species approach is limited to the range of the particular organism or organisms [94]. To understand, assess, and manage natural resources undergoing change requires a synthetic approach that looks across spatiotemporal scales and associated organizational levels [95,96]. Hierarchical classification systems for forests [97] and streams [98] have been specifically designed to allow integration of data for broader level assessments. However, forests are not always accurately described by an accounting of individual trees and large basins are not simply a sum of smaller catchments [99], and the presented habitat framework has the potential to effectively bridge these hierarchies of scale.