Wetlands

, Volume 30, Issue 2, pp 252–262

Does Long-term Grazing by Pack Stock in Subalpine Wet Meadows Result in Lasting Effects on Arthropod Assemblages?

Authors

    • University of California White Mountain Research Station
  • Jutta Schmidt-Gengenbach
    • University of California White Mountain Research Station
  • Sylvia A. Haultain
    • Sequoia and Kings Canyon National Parks
Original Paper

DOI: 10.1007/s13157-010-0020-3

Cite this article as:
Holmquist, J.G., Schmidt-Gengenbach, J. & Haultain, S.A. Wetlands (2010) 30: 252. doi:10.1007/s13157-010-0020-3

Abstract

Pack stock are often used in mountain environments and are grazed in uplands and wetlands, particularly subalpine wet meadows. Effects of pack stock on wetland invertebrates are unknown. Sequoia National Park, (Sierra Nevada, USA), was an ideal location for the study of lasting stock impacts on fauna, because a) there was an 18-year database of stock usage, b) there were meadows with little grazing that could be contrasted with grazed meadows, c) there is a long winter with no stock use, and d) the start of grazing for each meadow is controlled, so we could sample after greenup but just before stock arrived. We could thus address persistent conditions produced by many years of stock use in isolation from any potential short term impacts. We sampled terrestrial arthropods in paired “grazed” and “ungrazed” meadows across the Park and collected associated vegetation data. We found some negative effects of grazing on vegetation structure, but few lasting negative or positive effects of long-term stock grazing on arthropods in these wetlands. Although it appears that pack stock do not cause lasting damage to this arthropod assemblage, the extent of impact at the height of the grazing season remains unknown.

Keywords

AraneaeBaitingDisturbanceInsectaSierra Nevada (USA)Sweep netting

Introduction

Pack stock are frequently used on public lands in the Sierra Nevada (USA) and in other mountain environments, and these mules and horses, and occasionally burros and llamas, are often grazed in subalpine wet meadows (McClaran 1989; Spildie et al. 2000; Cole et al. 2004). The timing and locations for grazing are regulated in some areas (McClaran 1989; Moore et al. 2000; Spildie et al. 2000), but impacts to vegetation assemblages nevertheless can occur (Weaver and Dale 1978; Olson-Rutz et al. 1996a; Moore et al. 2000; Cole et al. 2004), and recovery is often not rapid (Olson-Rutz et al. 1996b; Spildie et al. 2000).

Although direct and indirect effects of outdoor recreation on invertebrates in vegetated assemblages have been demonstrated in a variety of ecosystems (e.g., Duffey 1975; Bayfield 1979; Eckrich and Holmquist 2000; Uhrin and Holmquist 2003), we are unaware of any studies on the effects of pack stock grazing on wetland invertebrates, despite the importance of invertebrates to the functional ecology of these habitats (van der Valk 2006; Williams 2006; Batzer and Sharitz 2006). Studies addressing use of grasslands, wetlands, and other vegetated habitats by different large mammals reveal mixed effects on the invertebrate assemblage (González-Megías et al. 2004; Underwood and Fisher 2006). Kruess and Tscharntke (2002) found cattle grazing to affect insects more than the plant assemblage, Bestelmeyer and Wiens (1996) recorded lower ant species richness as a function of cattle and goat grazing, and González-Megías et al. (2004) determined that sheep, goat, and ibex lowered diversity and abundance of arthropods in a Mediterranean mountain environment. Rambo and Faeth (1999) reported that deer, elk, and cattle reduced abundance, but not richness or evenness, of insects in a pine-grassland assemblage. Mysterud et al. (2005) found that sheep grazing in alpine pastures did not affect diversity or abundance of insects, and similarly Heske and Campbell (1991) and Bestelmeyer and Wiens (2001) discovered few differences in ant species richness, abundance, or assemblage structure as a function of livestock grazing. At the other end of the spectrum, Bock et al. (2006) found that grazing of small ranches by horses, cattle, and sheep can increase grasshopper abundances, and Majer and Beeston (1996) found higher ant species richness in more heavily grazed areas. Arthropod diversity can be increased by grazing via indirect effects mediated by shifts in canopy height, structural complexity, and plant diversity (Morris 1990; Olff and Ritchie 1998). Generalization concerning arthropod response to grazing across habitats is difficult due to the complex interactions of many factors (see literature surveys in González-Megías et al. 2004 and Underwood and Fisher 2006). Response of arthropods differs as a function of many variables, including livestock density, differences in grazing behavior, vegetation assemblage, duration of studies, arthropod response variables of interest, and especially duration of disturbance by livestock and the amount of time since last disturbance. Sierra Nevada wetlands were historically not grazed by herbivores larger than mule deer (Odocoileus hemionus Rafinesque, see Loomis et al. 1991; Loft et al. 1991; Dull 1999), and larger herbivores with different foraging behavior might be expected to cause shifts in both animal and plant assemblages.

If meadows that have been subject to long-term stock disturbance were to be sampled during a period of stock use, it would be difficult to determine if any apparent impacts were a function of long-term use, current use, or a combination thereof. In this study, we address one question regarding potential impacts of pack stock grazing on arthropod assemblages: Does grazing cause lasting effects that persist over time, or do long winters without stock allow an annual recovery of arthropod assemblages from any impacts that occur during summer usage? We do not address effects on wetland arthropod assemblages at the height of the grazing season. Sequoia National Park was an ideal location for this study, because a) there was a detailed, 18-year, meadow-specific database of stock usage, b) there were many meadows with little or no grazing use that could be contrasted with grazed meadows, c) there is a long winter period with no stock use, and d) the opening date for grazing on each meadow is controlled by the Park, so we could sample after greenup but just before stock arrived. Sampling prior to stock arrival allowed us to address lasting effects of many years of stock use in isolation from potentially confounding effects of current use.

Methods

We compared subalpine wet meadows with and without substantial stock use using a paired design. Generalization from responses of a single group of arthropods can be misleading (Gibson et al. 1992), so we examined effects across all canopy and ground-dwelling arthropod taxa that were collected by sweep nets and baits.

Study Area and Sites

Sequoia National Park is located in the southern Sierra Nevada mountains of California, USA (Fig. 1). The subalpine meadows that are open to pack stock are usually covered by snow for eight months, as are the 3000+ meter passes that allow access to these wetlands. Stock are not allowed into wetlands prior to snowmelt, and the National Park Service determines opening date for each meadow on the basis of the amount of snowfall for a given year, the timing of snowmelt, and the observed condition of the wetland in a given year. Access to selected wetlands is usually allowed about one month after snowmelt, i.e., individual wetlands are typically opened at various times from mid-June to mid-July. Stock use is intermittent from opening date through August, decreases rapidly after August, and ceases completely when the first substantial snow falls, generally in early November. Most stock use thus occurs over a two–three month period. Stock parties that used our study meadows ranged in size from one to 20 animals (\( \bar{x} = {8}.{2} \), SE = 0.68) in the five years preceding the study.
https://static-content.springer.com/image/art%3A10.1007%2Fs13157-010-0020-3/MediaObjects/13157_2010_20_Fig1_HTML.gif
Fig. 1

Sequoia National Park in the Sierra Nevada mountains in California, USA. Black circles represent grazed meadows paired with meadows with minimal stock use (white circles). Sites are separated slightly for clarity

Wet meadows are saturated with water much of the year (Williams 2006; Mitsch and Gosselink 2007), and these high diversity oases are common features of the subalpine environment (Körner 2003). The vegetation assemblages in the subalpine wet meadows managed for stock in Sequoia National Park are often dominated by a reed grass, Calamagrostis muiriana B.L. Wilson and S. Gray, formerly included in C. breweri Thurber, and we therefore focused our study on this assemblage (known as the Shorthair Reedgrass Herbaceous Alliance in the Sequoia/Kings Canyon National Park vegetation classification). Other important taxa include mountain ricegrass Ptilagrostis kingii (Bolander) Barkworth, tufted hairgrass Deschampsia cespitosa (L.) Beauv., various sedges Carex spp., rushes Juncus spp., shooting star Dodecatheon sp. (c.f. subalpinum Eastw.), club-moss Ivesia Ivesia lycopoidioides A. Gray, tundra aster Oreostemma alpigenum (Torr. and A. Gray) Greene, dwarf bilberry Vaccinium caespitosum Michaux, pussy-toes Antennaria spp., and western bistort Polygonum bistortoides Pursh. Klikoff (1965), Benedict (1983), and Ratliff (1985) provide good overviews of this assemblage. Assemblages characterized by higher levels of soil moisture, such as fens or wet meadows dominated by Deschampsia cespitosa and large sedges (e.g., Carex utriculata L. Bailey), are also grazed in the Sierra (Stohlgren et al. 1989).

We wished to contrast sites that had a long history of pack stock use with sites that had an equally long history of minimal use. Sequoia National Park has detailed records of pack stock use for many individual wetlands that span the last 18 years, as well as older, less formal records. We used these records to select 10 subalpine wet meadows (Table 1) that a) had been exposed to consistent use by pack stock (henceforth “grazed”), b) could each be paired with a subalpine wet meadow with little or no recent stock use (“ungrazed”), and c) were dominated by reed grass. Note that “grazed” in this context refers to all aspects of stock usage, including trampling, rather than cropping alone. Grazed and ungrazed conditions served as the mensurative treatments (Hurlbert 1984) for this study. We were able to locate pairs that were separated by an average of only 0.96 km (SE = 0.16) and 59 m (SE = 14) of elevation and that were in the same watersheds. We emphasized the tight co-location of paired grazed and ungrazed meadows in part to minimize geophysical and botanical differences. We also wished to minimize potential differences as a function of changing weather conditions, and the close proximity of each pair allowed us to sample both sites in rapid succession, before wind speed, air temperature, etc. could change greatly. Although each pair of wet meadows was tightly co-located, meadow blocks were separated by as much as 40 km, and up to two days of backpacking time was required to reach some sites. We thus sampled a relatively large number of blocks across a broad landscape with good replicate dispersion (Hurlbert 1984). We wanted sites to be as close to mid-season condition as possible in terms of the structure of the vegetation and arthropod assemblages, so we waited to sample until one hour to three days before stock reached the wet meadows in June and July of 2008.
Table 1

Site characteristics and usage patterns of grazed meadows over the past 18 years and over the past five years. All sites were wet meadows, regardless of place names. Each site was paired with a nearby ungrazed site (Fig. 1).

   

1990–2007

2003–2007

Site

Elevation (m)

Hectares

Total stock nights

Mean stock nights/year

Mean stock nights/ha/yr

Total stock nights

Mean stock nights/year

Mean stock nights/ha/yr

Hockett Pasture

2595

3.6

2590

144

39.5

408

82

22.4

South Fork Mdw

2587

4.5

3465

193

43.4

746

149

33.5

South Fork Pasture

2600

4.5

1387

77

17.3

195

39

8.8

Penned-Up Mdw

3242

5.3

621

35

6.6

199

40

7.6

Nathan’s Mdw

3061

5.9

1883

105

17.9

342

68

11.7

Lower Rock Ck Crossing

2893

25.5

2263

126

4.9

969

194

7.6

Lower Crabtree Mdw

3169

11.7

2042

113

9.6

585

117

10.0

Upper Crabtree Mdw

3192

17.0

2984

166

9.8

689

138

8.1

Tyndall Creek Mdw

3201

11.3

3352

186

16.4

711

142

12.6

Middle Rattlesnake Cyn

2907

5.3

1892

105

20.0

614

123

23.3

Mean

2945

9.5

2248

125

18.5

546

109

14.5

Std. Error

85

2.3

280

16

4.2

80

16

2.8

Each ungrazed or grazed site was sampled using a series of subsamples. We used aerial images of the sites to randomly select two 50 × 50 m subsample locations at each site prior to the field season. After arriving at each site, we established four additional sample locations, each at a randomly determined location within each of the pre-selected 50 × 50 m subsample locations. We used two of these sample locations for some metrics, and four for others (see below). Various metrics for a given ungrazed or grazed site were therefore means or composites of four or eight total measurements.

Field and Lab Methodology

Fauna

We used sweep nets to sample the meadow canopy fauna, and we supplemented these collections with baits targeting ground dwellers, particularly ants. Sweep nets are conical framed nets with a handle (New 1998; Southwood and Henderson 2000) and have a number of advantages for sampling remote areas. These nets are light in weight, easily transportable, do not impact wilderness character, integrate collections over a wide area, collect sparsely distributed species, can be used in habitats that are flooded or saturated with water, and produce samples that require relatively little sorting. Sweep nets have been shown to yield higher numbers of individuals, species, families, and orders, and capture higher levels of diversity than pitfall traps, light traps, or scented traps (Gadagkar et al. 1990). Sweep netting is probably the most widely used method for sampling arthropods in vegetation (Southwood and Henderson 2000), and this technique has been used in other investigations of the effects of grazers on arthropods (Rambo and Faeth 1999; Mysterud et al. 2005).

The response variables for each ungrazed or grazed site were means of two 50-sweep samples, with one 50-sweep sample from each of the 50 × 50 m subsample locations. We used a collapsible sweep net with a 30.5 cm aperture and mesh size of 0.5 × 0.75 mm (BioQuip #7112CP). Each of the two 50-sweep samples was in turn a composite of two 25-sweep subsamples from within each 50 × 50 m area. We sampled a total of 400 square meters at each site. Strengths of this approach include the previously noted integration of a large area and sampling of less common taxa, but conversely small scale invertebrate-habitat relationships (e.g., Crist et al. 1992; With 1994; Wiens et al. 1997) could have been missed. Sweeping was our first activity at each subsample location, because the subsequent work would have been likely to have disturbed fauna. Each sample was transferred to a self-sealing bag, killed with 99% ethyl acetate (Triplehorn and Johnson 2005), and kept as cool as possible until the trailhead was reached and the samples could be transferred to a freezer. All sweep sampling was done by a single worker throughout the project so as to minimize variance.

Baiting (Bestelmeyer et al. 2000; Delabie et al. 2000) targets ants and may also collect other taxa (Alonso 2000; Andersen and Majer 2004). Baiting is commonly used to monitor ant assemblages (Bestelmeyer et al. 2000) and has many of the same advantages as sweep nets for sampling remote areas. Our pilot tests of various bait combinations in 50 subalpine wet meadows over several years showed that honey and tuna baits offered the best combination of field practicality and attractiveness to multiple ant taxa. We placed one honey and one tuna bait within each of the two 50 × 50 m subsample locations at each site immediately after sweep netting, each at one of the 25-sweep locations. The baits consisted of ∼1 cm2 portions of honey or tuna and were placed on green construction paper cards and weighted by rocks. After 30 min, ants were removed with forceps and placed in a vial containing 70% ethanol. This method worked more reliably than preserving the entire bait or using an aspirator. The data from the honey and tuna baits at each subsample location were combined, and the data from the two subsample locations were used to generate means for response variables at each ungrazed or grazed site.

We sorted sweep samples in the lab, and identified taxa to family, with the exception of mites. Morphospecies counts were made for each sample. We identified ants from the bait samples to species.

Vegetation and Physical Data

We estimated percent green, standing brown (senescent), and litter cover as well as percent bare ground at the same two locations within each of the subsample locations that were used for sweep and bait samples. All of these metrics were visual cover estimates from a 10 × 10 m plot colocated with the area that was sweep netted. We measured canopy height and litter depth at two randomly selected locations within each area that was sweep netted. Cover estimates for each site were therefore means of four estimates, whereas canopy height and litter depth at each site were means of eight measures.

We recorded air temperature (in shade), relative humidity, and wind speed in the center of each 50 × 50 m subsample location using a Kestrel 3000 digital meter. These metrics were thus means of two measurements at each grazed or ungrazed site. Surface soil compaction was coarsely estimated with a Ben Meadows penetrometer at each of the canopy height/litter depth locations, thus yielding eight measurements per site.

Analysis

We performed 1 × 2 randomized block ANOVAs and ANCOVAs on a variety of faunal, vegetation, and physical metrics using SYSTAT 12. We analyzed a variety of faunal metrics, including order and family population abundances and family and morphospecies richness. Because large collections have more species than small collections, even if drawn from the same assemblage, we also assessed richness with expected number of species and families after scaling to the number of individuals in the sample with the fewest individuals (E(S8) and E(F8), Hurlbert 1971; Simberloff 1972; Magurran 2004). We analyzed family and morphospecies dominance and used probability of interspecific encounter, i.e., the probability that two species drawn from a sample are of different taxa, as a measure of evenness at both the morphospecies and family level (P.I.E., higher values indicate greater evenness, Hurlbert 1971). Margalef’s index (DMg, Clifford and Stephenson 1975; Magurran 2004) was used as a diversity measure for both families and morphospecies. We calculated E(S8), E(F8), and P.I.E. using the application Diversity. Some metrics demonstrated departures from normality via Lilliefors tests (Lilliefors 1967) and/or showed heteroscedasticity (Fmax and Cochran’s tests; Cochran 1941; Kirk 1982), but square-root transformations ((y)0.5 + (y + 1)0.5) of proportional data and log transformations (log (y + 1)) of all other data allowed parametric assumptions to be met. Only variables that differ as a function of treatment and that are not likely to be affected by the treatment should be considered for further analysis as covariates (Underwood 1997), and site elevation qualified via these criteria (Table 2). Although elevation differences between grazed-ungrazed pairs were small, most grazed sites were slightly lower than their associated ungrazed sites, and it was therefore important to examine elevation as a covariate. We present ANCOVA (general linear model) results for all response variables except air temperature, which was necessary to exclude because this variable did not meet the assumption of homogeneity of treatment and covariate regression slopes (Sokal and Rohlf 1995; Underwood 1997). Lastly, we constructed rank abundance plots which provide an additional perspective on diversity, richness, and evenness, without collapsing a great deal of information into a single number (Stiling 2001; Magurran 2004; Underwood and Fisher 2006).
Table 2

Means, standard errors, and results of 1 × 2 randomized block ANOVAs (n = 20; df = 1,9) and ANCOVAs with elevation as a covariate (df = 1,9,1) for vegetation and physical metrics. No ANCOVA for air temperature due to heterogeneity of treatment and covariate slopes

 

Ungrazed

Grazed

ANOVA

ANCOVA

Mean

SE

Mean

SE

Block

Treatment

Block

Treatment

Elevation (m)

3013.9

90.5

2954.1

85.6

<0.01**

<0.01**

NA

Canopy height (cm)

8.9

0.9

8.5

0.9

0.19

0.73

0.32

0.44

Litter depth (cm)

1.9

0.2

0.8

0.2

0.14*

<0.01**

0.18*

0.08*

Litter cover (%)

4.6

0.8

2.4

1.0

0.09*

0.02**

0.10*

0.12*

Bare ground (%)

7.8

2.9

12.6

3.7

0.91

0.33

0.95

0.55

Brown cover (%)

6.4

1.7

2.3

0.7

0.16*

0.03**

0.21

0.26

Green cover (%)

81.2

4.0

82.7

3.8

0.84

0.82

0.91

0.90

Wind speed (km/hr)

6.3

0.6

5.7

0.6

0.67

0.50

0.58

0.23

Air temperature (°C)

20.2

0.8

20.0

1.0

<0.01**

0.64

NA

 

Humidity (%)

32.1

2.8

33.7

1.7

0.11*

0.39

0.20

0.72

Soil compaction (kg/cm2)

1.5

0.1

1.5

0.2

0.75

0.98

0.83

0.67

*P < 0.19 (see Methods); **P < 0.05.

Because this study addressed potential anthropogenic impacts, we wanted good power and tight control over Type II error. Although ecologists tend to emphasize Type I error over Type II error, there is often not an ecological basis for this bias, particularly in situations that involve potential environmental degradation. It is increasingly recognized that both types of error deserve equal scrutiny, and it can be advantageous to set alpha as high or even higher than beta in order to increase power and decrease Type II error (Kendall et al. 1992; Mapstone 1995; Dayton 1998; Field et al. 2004). We used as many replicate ungrazed blocks (10 wet meadow pairs) as possible; we were not able to use more sites because of the limited number of subalpine wet meadows that met our criteria for pairing, so power could not be increased by increasing sample size. Before conducting our field work, we used G*Power (Erdfelder et al. 1996; Faul et al. 2007; Mayr et al. 2007), our known sampling design and sample size, and the standard a priori estimate for effect size of 0.5, which has been well-established both theoretically and empirically though large meta-analyses (Cohen 1988; Lipsey and Wilson 1993; Bausell and Li 2002) to estimate the a priori alpha level that would be required in order to have an equivalent beta error. The result was alpha = beta = 0.19, and the associated power (1-beta) was 0.81. Note that this is not retrospective power analysis, which is not recommended (Hoenig and Heisy 2001; Nakagawa and Foster 2004). In contrast, the a priori beta estimate using these same parameters for a fixed alpha of 0.05 was 0.44 and a power of only 0.56, which would give good protection from Type I error, but poor protection from Type II error and therefore a greater chance of falsely assuming that pack stock have little effect on wet meadow arthropods. We used both alpha = 0.19 and the standard alpha = 0.05 as significance thresholds in order to provide additional perspective for our results.

Results

Both sets of meadows had > 80% green vegetation, ∼8 cm canopy height, equal soil compaction, and similar percent bare ground, but there were some differences in vegetation structure (Table 2). Grazed meadows had shallower litter depth as well as lower percent litter and brown vegetation cover (ANOVA). Some differences among meadow pairs (block effects) were apparent for these variables as well as for temperature and humidity. Effects were generally lessened when analyzed via ANCOVA, although both litter depth and cover were still different (Table 2) at alpha = 0.19 (see Methods).

We collected and identified 2,683 arthropods in the study, representing 11 orders and 81 families. Diptera had the greatest family richness (29), followed by Hemiptera (12), Hymenoptera (12), and Coleoptera (10). There were 68 families in the ungrazed samples and 63 families in the grazed samples. Rank abundance plots for the two meadow conditions were similar (Fig. 2), and both plots fell between log normal and broken stick configurations. There was slightly more abundance at family ranks 7 through 20 on the grazed plots and slightly more abundance at ranks 20 through 43 on the ungrazed plots.
https://static-content.springer.com/image/art%3A10.1007%2Fs13157-010-0020-3/MediaObjects/13157_2010_20_Fig2_HTML.gif
Fig. 2

Rank abundance plot for families comparing ungrazed and grazed wet meadows based on total abundances for study

Sweep assemblage metrics for ungrazed and grazed meadows were almost identical when assessed via ANOVA or ANCOVA (Table 3). There was also little evidence of block effects at the assemblage level. The overall assemblage was dominated by Diptera and Hemiptera at the order level (Table 4). Three of the four most abundant families (and 7 of the top 10) were dipterans; anthomyiid flies had the highest overall family abundance, followed by cicadellid leafhoppers, ephydrid shore flies, and chloropid grass flies. Ungrazed and grazed plots had the same six most abundant families, although the rank order differed. Only Diptera and Hemiptera were found in all samples; at the family level, chloropids, muscid house flies, and anthomyiids were found in almost all samples (Table 4). Coleoptera was the only group that differed in abundance between ungrazed and grazed meadows when assessed with ANOVA, indicating more beetles on ungrazed sites, although beetles were relatively uncommon in the assemblage. ANCOVAs that included elevation as a covariate similarly did not reveal grazed-ungrazed differences for abundant taxa, but did show larger numbers of Orthoptera, fungus gnats (Sciaridae), and spiders (Araneae) on grazed sites. Approximately one-third of the population variables had significant block effects, the strongest of which were for Orthoptera, Sciaridae, Anthomyiidae, and Araneae (Table 4).
Table 3

Means, standard errors, and results of 1 × 2 randomized block ANOVAs (n = 20; df = 1,9) and ANCOVAs with elevation as a covariate (df = 1,9,1) for assemblage-level faunal metrics. All metrics are based on 50-sweep samples

 

Ungrazed

Grazed

ANOVA

ANCOVA

Mean

SE

Mean

SE

Block

Treatment

Block

Treatment

Total individuals

67.6

15.98

66.6

14.60

0.28

0.90

0.32

0.73

Family richness

13.6

1.91

13.4

1.50

0.61

0.91

0.65

0.61

Species richness

18.3

2.87

18.2

2.20

0.55

0.83

0.63

0.78

Expected no. of families E(F8)

4.8

0.33

4.8

0.15

0.26

0.62

0.25

0.58

Expected no. of species E(S8)

5.3

0.39

5.5

0.24

0.12*

0.45

0.17*

0.77

% Family dominance

40.1

4.72

37.1

2.63

0.23

0.62

0.23

0.59

% Species dominance

35.8

5.48

30.1

3.66

0.26

0.39

0.34

0.85

Hurlbert’s PIE (family)

0.8

0.04

0.8

0.02

0.35

0.45

0.32

0.67

Hurlbert’s PIE (species)

0.8

0.05

0.8

0.02

0.27

0.36

0.28

0.73

Margalef’s family diversity

3.2

0.36

3.0

0.22

0.43

0.75

0.56

0.53

Margalef’s species diversity

4.3

0.56

4.3

0.36

0.41

0.86

0.53

0.78

% Predators

11.3

2.04

12.0

2.31

0.12*

0.82

0.15*

0.69

*P < 0.19 (see Methods); **P < 0.05.

Table 4

Means, standard errors, frequencies, and results of 1 × 2 randomized block ANOVAs (n = 20; df = 1,9) and ANCOVAs with elevation as a covariate (df = 1,9,1) on abundances of orders and the ten most abundant families in sweep samples (top) and on bait metrics (below). Sweep metrics are based on 50-sweep samples, and all bait metrics are per one aggregate hour of bait deployment using one honey and one tuna bait. Plecoptera and Psocoptera were too rare to test

 

Ungrazed

Grazed

ANOVA

ANCOVA

Mean

SE

Frequency

Mean

SE

Frequency

Block

Treatment

Block

Treatment

Sweeps

 Orthoptera

0.45

0.29

0.40

1.20

0.76

0.40

0.04**

0.31

0.12*

0.11*

 Plecoptera

0.05

0.05

0.10

0.00

0.00

0.00

NA

NA

 Hemiptera

18.70

6.16

1.00

15.55

4.12

1.00

0.11*

0.97

0.24

0.88

 Cicadellidae

9.10

3.97

1.00

9.70

3.97

0.70

0.27

0.87

0.75

0.32

 Delphacidae

7.25

4.83

0.80

4.40

1.31

0.90

0.68

0.77

0.52

0.20

 Thysanoptera

0.20

0.11

0.30

0.10

0.07

0.20

0.81

0.54

0.90

0.83

 Psocoptera

0.00

0.00

0.00

0.05

0.05

0.10

NA

NA

 Coleoptera

1.05

0.23

0.80

0.40

0.12

0.60

0.95

0.10*

0.91

0.80

 Hymenoptera

2.90

0.66

0.90

3.60

1.34

1.00

0.17*

0.76

0.16*

0.39

 Ichneumonidae

0.90

0.34

0.70

1.70

1.00

0.70

0.07*

0.49

0.07*

0.58

 Lepidoptera

0.15

0.08

0.30

0.30

0.20

0.30

0.82

0.65

0.52

0.28

 Diptera

43.20

14.38

1.00

41.95

10.20

1.00

0.42

0.93

0.37

0.42

 Culicidae

1.40

0.97

0.50

1.85

1.37

0.50

0.05*

0.85

0.49

0.54

 Sciaridae

1.15

0.43

0.60

1.55

1.09

0.50

0.11*

0.72

<0.01**

<0.01**

 Empididae

1.50

1.28

0.50

1.15

0.49

0.50

0.22

0.64

0.21

0.29

 Anthomyiidae

7.35

2.94

1.00

11.70

5.42

0.90

<0.01**

0.50

<0.01**

0.57

 Muscidae

6.55

2.25

0.90

6.80

1.68

1.00

0.83

0.49

0.54

0.28

 Chloropidae

6.10

1.75

0.90

10.20

3.26

1.00

0.91

0.38

0.96

0.53

 Ephydridae

12.90

11.52

0.90

4.15

1.94

0.60

0.26

0.94

0.39

0.72

 Araneae

0.85

0.30

0.70

1.05

0.73

0.50

0.07*

0.69

0.01**

0.03**

 Acari

0.15

0.11

0.20

0.30

0.25

0.20

0.39

0.70

0.54

0.78

Baits

 Myrmica discontinua

1.55

0.93

0.60

3.40

1.81

0.30

0.61

0.75

0.30

0.20

 Formicidae

1.95

0.95

0.70

4.05

1.82

0.50

0.44

0.58

0.12*

0.14*

 Ant species richness

0.55

0.19

 

0.55

0.20

 

0.34

0.90

0.11*

0.81

 Acari

0.25

0.13

0.30

0.80

0.64

0.30

0.37

0.55

0.46

0.22

*P < 0.19 (see Methods); **P < 0.05.

Relatively few taxa and individuals were collected on the bait cards (Table 4). The ant (Formicidae) catch was dominated by Myrmica discontinua Weber, but we also collected small numbers of Formica lasioides Emery, F. neorufibarbis Emery, F. aserva Forel, F. canadensis Santschi, and Camponotus vicinus Mayr, as well as Acari (mites). Species richness was identical in grazed and ungrazed meadows, but total ant abundance was twice as high on grazed as on ungrazed sites (p = 0.14, ANCOVA, Table 4). ANCOVAs also showed significant block effects for ant abundance and species richness. No significant differences were apparent via ANOVA.

Discussion

Some changes in coarse vegetation structure persisted from previous years of stock use to the start of the new grazing season, despite the annual winter respite from stock use. There was significantly less litter depth and cover on our grazed sites, and reductions in litter have also been observed as a result of cattle and sheep grazing (King and Hutchinson 1983; Andresen et al. 1990; Bromham et al. 1999). We observed less standing, senescent (brown) vegetation at the grazed sites, and this effect was probably due to breakage and grazing of vegetation at the end of the previous year. Although bare ground was nominally more extensive on our grazed sites, this difference was not significant, in contrast with findings from past manipulations of pack stock use (Moore et al. 2000; Spildie et al. 2000; Cole et al. 2004). We also found no evidence of lasting impacts on canopy height, in contrast to several other studies of livestock effects (Andresen et al. 1990; Kruess and Tscharntke 2002; Hartley et al. 2003). This result is important, because canopy height is often a positive predictor of insect diversity and abundance (Haysom and Coulson 1998; Kruess and Tscharntke 2002). Experimental clipping (Stohlgren et al. 1989) indicates that wetter vegetation assemblages may be more susceptible to livestock impact than the Calamagrostis dominated assemblage.

We found relatively few negative or positive effects of long-term pack stock grazing on the arthropod assemblages in these subalpine wet meadows, but we addressed only persisting multi-year effects rather than the immediate effects that may occur at the height of stock usage. Hatfield and LeBuhn (2007) found sheep grazing to negatively affect bumble bee assemblages in the Sierra Nevada but similarly found these effects to not carry over to a subsequent year. Our one significant faunal contrast via ANOVA showed beetles to have a negative response to grazing, whereas ANCOVA showed positive effects on four taxa, including ants. Studies of livestock effects on arthropods have variously found positive, negative, mixed, or no effect across the entire assemblage (see Introduction); other efforts report differential responses among arthropod taxa. Herbivores (Andresen et al. 1990; Gibson et al. 1992) and leafhoppers in particular, have been found to be more affected by livestock than other taxa (Morris and Lakhani 1979; Morris and Rispin 1987; but see Kruess and Tscharntke 2002). Although invertebrates have been shown to be notoriously sensitive to subtle vegetation differences in many environments (e.g., Wiens et al. 1997; Holmquist 1998; McAbendroth et al. 2005), there were apparently few indirect effects on arthropods driven by litter losses in the grazed meadows. Ants might represent an exception. This group showed a significant positive relationship to grazing, albeit at a higher alpha level and only via ANCOVA after adjusting for elevation. Ants have been shown to have positive responses to livestock grazing in some other habitats (e.g., Majer and Beeston 1996; Bromham et al. 1999; Underwood and Fisher 2006), and these increases can be driven by litter losses (Bromham et al. 1999) similar to those observed in our study.

Were there really few effects on fauna? The almost complete lack of significant negative effects on fauna as tested by ANOVA across 12 assemblage and 21 population metrics, not only at alpha = 0.05 but at the high alpha of 0.19 and associated high power of 0.81, provides no indication of an overall negative grazing effect on fauna. The rank abundance plots were also consistent with this conclusion. Analysis by ANCOVA also did not suggest negative effects on faunal assemblage metrics or populations, or positive effects on assemblage metrics, but did suggest positive effects in four of 21 tested faunal populations (orthopterans, fungus gnats, spiders, and ants). These positive effects may be in fact be present, and other work has shown both orthopterans (Bock et al. 2006) and ants (Underwood and Fisher 2006) to be positively affected by grazing in some habitats. It is also possible that, given the small elevation differences between grazed and ungrazed treatments, the statistical significance of elevation may exceed the associated ecological significance. The presence of significant block effects for about one-third of the faunal population, physical, and vegetation metrics indicates that there were some differences among habitats and that there was sufficient power in the design to detect extensive treatment differences if such differences were present. Although negative effects were generally not observed at the family level, it is possible that some individual species were reduced in abundance or absent on the grazed sites.

Why were there no negative effects on fauna? Negative livestock effects on fauna have been demonstrated in a number of environments (e.g., Bestelmeyer and Wiens 1996; González-Megías et al. 2004), and Kruess and Tscharntke (2002) found insects to be more sensitive than plants to cattle grazing. Although many studies of grazing effects on arthropods have used spatial comparisons, several authors have shown arthropod population densities, biomass, species richness, and/or diversity to increase when stock pressure ceases (Andresen et al 1990 and references therein; Hatfield and LeBuhn 2007). Sequoia National Park, however, has a particularly rigorous stock management program and strives to limit pack stock impacts by controlling opening dates for individual wetlands on the basis of wet meadow condition, assessed via plant assemblage structure and phenological development, as well as estimates of soil moisture determined by the nature of the preceding winter. Although we found some stock impacts on dead vegetation structure, the low levels of stock usage maintained by Sequoia National Park (mean of 18.5 stock nights/ha/yr) were apparently below the threshold for impact to the arthropod assemblage, at least as assessed at the start of the growing season before stock arrived. Cole et al. (2004) note that meadow vegetation can be maintained in good condition with low levels of stock use, but even moderate use often results in impacts. Park Service regulation of meadow opening dates and stock densities, arthropod dispersal capabilities (Hatfield and LeBuhn 2007), concentration of grazing in Calamagrostis dominated meadows, short grazing seasons, long winter recovery periods, and our pre-grazing sampling likely combined to limit impacts and/or our detection thereof. There might be different results in Sierra meadows with wetter conditions (e.g., Stohlgren et al. 1989), a longer grazing season, or less regulation. As an example of one impact pathway that was absent in our wetlands, Andresen et al. (1990) found cattle to reduce canopy height with an associated loss of canopy arthropods. In our Sequoia wet meadows, stock were excluded from grazing areas in the spring and a full canopy developed. We sampled before stock arrived and thus before the canopy could potentially be newly degraded. Our results indicate little long-term stock damage to the arthropod assemblage, and this finding is encouraging, but our results do not address potential impacts at mid-season.

If there were no lasting negative effects on fauna, does that mean that any mid-season impacts are inconsequential? No. The many studies recording livestock impacts to epigeal arthropods report results obtained during or immediately after grazing (Kruess and Tscharntke 2002; see Introduction). Our limited mid-season pilot sampling also suggests that pack stock may reduce arthropod diversity and abundance in these subalpine wet meadows. Removal of canopy in the middle of the growing season is more deleterious to the arthropod assemblage than removal during early season (Duffey et al. 1974), and Baines et al. (1998) showed that one mid-season canopy removal had greater negative effect on spider species richness and abundance than two removals in spring and fall. Univoltine species can be affected by mid-season canopy removal more than multivoltine species (Morris 1979), but multivoltine taxa could also have brood size reduced or eliminated during peak stock use periods. Flowering in these subalpine wetlands occurs during stock usage, and removal of flowers can negatively affect butterflies (Feber et al. 1996), and other nectivores (Vickery et al. 2001; Hatfield and LeBuhn 2007), which in turn may reduce pollinator availability to plants. These subalpine wetlands may be “reset” over the long winter and spring, but it is possible that pack stock reduce mid-season productivity and diversity of these wetlands, and such losses could cascade into vertebrate (Vickery et al. 2001) and/or upland assemblages. Although it appears that pack stock do not cause lasting damage to the wet meadow arthropod assemblage, the question as to impacts at the height of the grazing season remains unanswered, and we will not have a full understanding of the role of pack stock in these wetlands until this issue is addressed.

Acknowledgements

We thank Lyra Pierotti, Chelsea Clifford, Jean Dillingham, Derham Giuliani, and Peter Norquist for cheerfully sorting samples and Eric Frenzel for helping compile historical stock use data. Philip Ward kindly confirmed ant species identifications. We benefited from discussion with Harold Werner, Leigh Ann Starcevich, Liz Ballenger, Peggy Moore, David Graber, Jennifer Jones, David Cooper, Eric Berlow, and Linda Mutch, and from the support of WMRS faculty and staff, especially Vikki DeVries, Barbara Fager, Frank Powell, Daniel Pritchett, Elizabeth Sally, John Smiley, and Denise Waterbury. The paper was improved by comments from Darold Batzer, Scott Martens, Peggy Moore, Steve Ostoja, John Smiley, and anonymous reviewers. This work was funded by the National Park Service (J8R07080005 and J8R07070006). Much of the groundwork for this project was supported by the National Park Service (H8R07010001) and National Science Foundation (0139633 and 0139633-Supplement). All Park Service support was funded through the Great Basin Cooperative Ecosystems Studies Unit, and Angie Evenden expertly assisted with this process.

Copyright information

© Society of Wetland Scientists 2010