MONITORING SPRING-SUPPORTED VEGETATION ...

3 downloads 0 Views 1MB Size Report
John R. Spence. Science & Resource Management. Glen Canyon NRA. 17 July 2013. Limnocrene spring and associated pool in Coyote Gulch, Glen Canyon ...
MONITORING SPRING-SUPPORTED VEGETATION IN SOUTHWESTERN PARKS: CONCEPTUAL ISSUES AND APPROACHES

John R. Spence Science & Resource Management Glen Canyon NRA 17 July 2013

Limnocrene spring and associated pool in Coyote Gulch, Glen Canyon NRA (J. Spence NPS Photo)

I. INTRODUCTION There are a wide variety of methods for monitoring spring and other ground-water dependent plant communities and wetland species, ranging from more complex and detailed quantitative sampling, to remotely-sensed methods (ie., LiDAR, high resolution photography), or simple species presence-absence (incidence monitoring). This report looks at issues and approaches associated with vegetation and plant species monitoring at springs for the NPS Inventory & Monitoring Program, using examples and data drawn primarily from southwestern springs of the Sonoran and Colorado Plateau regions. It should be considered a preliminary assessment of various methods that could be applied to monitor vegetation associated with ground-water dependent systems. Approaches can be grouped into two main categories of monitoring the status of vegetation and plants on a landscape over time. For many resources of interest, detailed (intensive) repeated sampling of selected representative sites may be required to detect change. This approach has been adopted by most I&M Networks for monitoring, e.g., for upland and riparian vegetation. An alternative approach is to conduct extensive monitoring where less site-specific detail is needed or required, and a larger sample size can be achieved (Philippi et al. 1998). This second approach may be of use for resources or sample sites that are easily damaged or disturbed by repeat visits, such as sensitive small-scale springs and tinajas. Repeated intensive sampling of vegetation at a spring can potentially produce changes in vegetation resulting from the visit itself, which in turn can mask background change in those factors that the monitoring is attempting to detect, such as effects of long-term climate change, ground water depletion, or species extinction and colonization. In this report I examine general features of several different monitoring methods suitable for detecting change over time for both spring-supported vegetation and individual plant species at ground-water dependent systems in the Southwest. Various approaches are suggested, with strengths and weaknesses of each evaluated. Significantly more discussion on incidence (occupancy) monitoring is included as this set of methods has not typically been applied to plant communities. First, a general discussion of some scale and sampling ideas is introduced, primarily on how to partition groups of springs based on various geographic attributes. Then six general monitoring approaches are presented and sketched out in general detail. Finally, some recommendations are made based on logistics and monitoring requirements. II. SAMPLING AND SCALE CONSIDERATIONS Experimental design and sampling theory are often different between the two approaches discussed in the Introduction. Intensive sampling of sites includes variability and errors associated with measures of species performance, e.g., abundance or canopy cover. An extensive approach that looks at species presence over time (occupancy or ψ) across a sample of sites introduces at least one additional sources of error. This is under-sampling or sampling error; or how many species are identified at a site in a single visit. This can be partitioned into two variables, observer error, and under sampling error (also known as detectability) due to the presence of smaller obscure species, species with differing phenologies, or failing to find a species in large complex sites. Another general source of error consists of natural background 2

temporal variability (not directional change or “trend”; e.g., Urquhart et al. 1998) resulting from chance factors, such as disturbance leading to local extirpation or invasion, dispersal, etc. As pointed out elsewhere (e.g., Noon et al. 2012) the variable of interest in occupancy modeling would be the overall change in occupancy among sites over time, or ∆ψ. These ideas are explored more thoroughly below under Occupancy and Probability Detection Theory (see also Table 1). Springs vary in elevation, disturbance history, size and other features. There is also significant beta diversity (species turnover) among springs, even when other factors are controlled, due to chance extinction and immigration events. Thus some form of preliminary segregation of springs is required to partition out those factors that can be controlled for. Otherwise the inherent variability in vegetation or species present at sites, if large enough, can potentially mask more subtle long-term trends. Figure 1 illustrates a small watershed with a variety of springs, including rheocrene, hanging garden and hillslope springs. Rheocrene spring are likely to have much higher levels of disturbance due to in-stream and usually annual flooding, thus higher turnover of species compared to, e.g., more stable hanging gardens in alcoves or off-channel settings. In an example like this, springs should be grouped by similarity among various significant controlling factors. Thus all rheocrene springs could be grouped together, or perhaps partitioned again by elevation if significant floristic differences occur. These groups could then be aggregated at the larger geographic scope such as Sonoran Desert parks to derive a sufficiently large sample size. Table 1 summarizes some of the main spring types in the Southwest and provides a preliminary attempt to identify some of the principal disturbance factors.

Table 1. List of principal spring types, geomorphological locations, and common disturbances. Other types of springs are either very rare, or cannot typically be monitored for vegetation. Spring classification from Springer and Stevens (2009) with modifications. Drought effects on aquifers and re-charge are common to all spring types. Spring Type Geomorphic Setting Principal Physical Disturbances Gushet Spring Hanging Garden

Variable, in-channel or offchannel Variable, in-channel or offchannel

Hillslope Spring

Off-channel on slopes

Heliocrene Spring

Variable, in-channel or offchannel, low gradient Variable, in-channel or offchannel

Hypocrene Spring Rheocrene Spring

In stream channels

Tinaja Springs1

Bedrock controlled intermittent or ephemeral stream channels; water flows into rock basin Highly variable; sediment-filled plunge pools with gravity flow

Plunge-pool Springs2 1a

Rock falls; flooding if associated with stream channels Rock falls; including spalling in alcoves; slope failures; flooding (in-channel); plunge pool flooding if in theater- network canyons Surface disturbances affecting erosion and compaction processes; slope failures; rock falls; overland (sheet) flooding Surface disturbances affecting erosion and compaction processes Surface disturbances affecting erosion and compaction processes; slope failures; rock falls; overland (sheet) flooding In-channel flooding along with associated debris and sediment movements; slope failures In-channel flooding along with associated debris and sediment movements Flooding, head cutting; surface disturbances affecting erosion and compaction processes

special subset of rheocrene springs; 2see Spence (2013)

3

Divide Watershed Boundary HS HS HS Upper Watershed

RS HG HG RS RS HG RS

Middle Watershed

Elevation

HS Distance from Divide

HG RS HS HS Lower Watershed RS

HG

Figure 1. Hypothetical watershed characterized by several different spring systems, including rheocrene springs (RS), hillslope springs (HS) and hanging gardens (HG). The spring classification is from Springer and Stevens (2009).

4

III. POTENTIAL MONITORING METHODS Six general widely used approaches to monitoring spring vegetation and species are described below. Four of these approaches are based on standard well known ecological sampling methods, including cover estimates or other standard measures of species performance, use of transects or quadrats, and demographic monitoring of selected species. The fifth approach consists of standard photo point monitoring. The final approach includes a more detailed discussion and analysis of species incidence monitoring, which has generally not been applied to plants in this context, and thus requires more discussion. 1. Wetted Areal Extent In this approach, the wetland vegetation (general wetted area) supporting obligate wetland plant species (or other surrogate such as wetted soils) is delineated either with GPS or by fixing permanent points and measuring cross-sections or areas. The illustration (Figure 2) shows a small wetland supported by a spring with five permanently located transects with lengths measured. Repeat visits can determine if the overall wetted area is shrinking or expanding. When satellites are easily accessible, a GPS unit with centimeter-level precision can also be used to map out the entire polygon. In this latter case change is analyzed as: A1 = A0 + ΔA(0→1) Where A = wetted area (m2, hectares, etc.) at times 0 and 1 (subsequent repeated sample). Limitations of this method are that it will not work well for highly irregular or dissected wetted areas, and for tinajas in bedrock depressions. In addition, no species composition data is collected. The principal advantage is that this approach is simple and quick to complete, and can be directly correlated with changes in spring discharge.

Figure 2. Sample plan view of a wetland area supported by spring.

2. Species Cover and Vegetation Composition This is one of the more common protocols used by various I&M Networks, as well as the US Forest Service (2012), for monitoring work on spring vegetation. The spring environment is first mapped or sketched, and distinct geomorphic units are then delineated. On each unit the presence and cover of species is listed, typically using visual-only estimates (but see below). 5

Figure 3 shows a small classic alcove hanging garden with three distinct units, a sloping wetted backwall, a detritus slope, and a wetland/wet meadow downslope. Species lists for each of these units are made, and approximate canopy cover estimated using one of the standard scales listed in Table 2.

Figure 3. Sample plan view of a simple alcove hanging garden with three wetted geomorphic surfaces. Backwall

Detritus Slope

Wet meadow

Stream

A variant of this approach is to subsample the geomorphic units using either smaller quadrats or placement of permanent transects. Quadrats can either be randomly placed in the geomorphic unit, or systematically placed along transects (e.g., Spence and Henderson 1993; Fowler et al. 2007). Another method is to use the point count approach, where a pole is placed along transects at pre-determined points, with those species touching the pole recorded. Considerations of adequate sample and quadrat sizes are needed for these methods, which are not discussed here but can be found in standard plant ecology books (e.g., Shimwell 1971; Mueller-Dombois and Ellenberg 1973; Grieg-Smith 1983; Bonham 1989; Elzinga et al. 1998). Change is determined by either total canopy cover or by individual species cover (A attributes) as: AGUx (time 1) = AGUx (time 0) + ΔAGUx (time 0→1) AGUy (time 1) = AGUy (time 0) + ΔAGUy (time 0→1) AGUz (time 1) = AGUz (time 0) + ΔAGUy (time 0→1) Where A = attribute being measured, and GUx-y = geomorphic units x through z. Among the advantages of these methods is that they provide relatively complete species lists, and estimates of canopy cover varying in precision and accuracy estimates. The principal disadvantages include significant time costs, and potentially extensive damage to the spring vegetation from trampling. In this context, the first method listed using a general geomorphic unit and general cover estimate is less time-consuming and less likely to disturb sites than more intensive sampling approaches using transects and quadrats. 6

Table 2. Three general cover canopy scales that can be used for visual cover estimates of vegetation at springs. Modified BraunBlanquet Cover Class 10 9 8 7 6 5 4

Canopy Cover 90-100% 80-90% 70-80% 60-70% 50-60% 40-50% 30-40%

MidPoint 95 85 75 65 55 45 35

3

20-30%

25

2

10-20%

15

1

1-10%

5

++