Journal of Vegetation Science doi: 10.3170/2008-8-18454, published online 15 April 2008 © IAVS; Opulus Press - AUppsala. simple classification of biological

soil crust habitat on the Colorado Plateau -

1

A simple classification of soil types as habitats of biological soil crusts on the Colorado Plateau, USA Bowker, Matthew A.1,2* & Belnap, Jayne3 1Department

of Biological Sciences, Northern Arizona University, Box 5640, Flagstaff, AZ 86011,USA; Current Address: Área de Biodiversidad y Conservación, Universidad Rey Juan Carlos, c/ Tulipán s/n., E-28933 Móstoles (Madrid), Spain;3Southwest Biological Science Center, U.S. Geological Survey, 2290 S.W. Resource Boulevard, Moab, UT 84532, USA; E-mail: [email protected]; *Corresponding author; E-mail [email protected] Abstract Question: Can a simple soil classification method, accessible to non-experts, be used to infer properties of the biological soil crust (BSC) communities such as species richness, evenness, and structure? Location: Grand Staircase-Escalante National Monument, an arid region of the Colorado Plateau, USA. Methods: Biological soil crusts are highly functional soil surface communities of mosses, lichens and cyanobacteria that are vulnerable to soil surface disturbances such as grazing. We sampled BSC communities at 114 relatively undisturbed sites. We developed an eight-tier BSC habitat classification based upon soil properties including texture, carbonate and gypsum content, and presence of shrinking-swelling clays. We used simple structural equation models to determine how well this classification system predicted the evenness, richness, and community structure of BSC relative to elevation and annual precipitation. Results: We found that our habitat classification system explained at least 3.5 × more variance in BSC richness (R2 = 0.57), evenness (R2 = 0.59), and community structure (R2 = 0.34) than annual precipitation and elevation combined. Gypsiferous soils, non-calcareous sandy soils, and limestonederived soils were all very high in both species richness and evenness. Additionally, we found that gypsiferous soils were the most biologically unique group, harboring eight strong to excellent indicator species. Conclusions: Community properties of BSCs are overwhelmingly influenced by edaphic factors. These factors can be summarized efficiently by land managers and laypeople using a simple soil habitat classification, which will facilitate incorporation of BSCs into assessment and monitoring protocols and help prioritize conservation or restoration efforts. Keywords: Biodiversity conservation; Cryptogam; Ecological indicator; Ecosystem engineer; Land management; Microbiotic crust; Soil chemistry. Nomenclature: Zander 2008 (Bryophytes); Esslinger 2008 (Lichen). Abbreviation: BSC = Biological soil crust.

Introduction Biological soil crusts (BSCs) are a type of thin (< 1 cm), desiccation tolerant microbial mat of cyanobacteria, subsequently colonized by mosses and lichens, living at the soil surface in drylands. These BSCs act as ecosystem engineers (Jones et al. 1997) and contribute strongly to dryland ecosystem function (Belnap 2002; Beymer & Klopatek 1991). For example, the filamentous cyanobacteria of BSCs chemically and physically aggregate the soil surface into a thin horizontal layer, increasing erosion resistance and influencing water redistribution (Mazor et al. 1996). In arid lands, vascular plants are known to create ‘fertility islands’, but BSCs are a dominant influence in the creation and maintenance of soil fertility in interspaces between vascular plants. Mechanisms through which BSCs increase fertility include carbon (C)-fixation (Beymer & Klopatek 1991), nitrogen (N)fixation (Belnap 2002), and dust trapping (Reynolds et al. 2000), among others. Biological soil crusts also have potential to be highly valuable as ecological indicators of ecosystem health (Tongway & Hindley 1996). For these reasons and others, scientists should enable the inclusion of these information-rich ecological indicators in the management of public lands and other commons, in addition to private lands. Many previous studies have related environmental factors to the community composition and structure of BSCs at both small (< 1 m; Bowker et al. 2006a; Rosentreter 1986; Martinez et al. 2006) and large scales (100s of km2; Ponzetti & McCune 2001; Bowker et al. 2005; Thompson et al. 2006). Often, a confusing array of inter-correlated factors also correlate with specific components of the BSC community (Eldridge & Tozer 1997). Not surprisingly, BSC abundance, composition, and distribution are influenced by climate variables such as annual precipitation, maximal temperature, and proximity to maritime air (Rogers 1972a; Nash & Moser 1982). Another strong influence is the physical and chemical environment of the soil (Downing & Selkirk 1993; Ponzetti­ & McCune 2001; Bowker et al. 2005).

2

Bowker, M.A. & Belnap, J

Soil chemical and physical properties are especially difficult to consider independently, for example pH, CaCO3, electrical conductivity, and available fractions of immobile nutrients (P, Mn, Zn, Cu, Fe) are usually strongly autocorrelated. Surface disturbances, such as livestock grazing, are also highly influential in reducing BSC abundance (Brotherson et al. 1983). Potentially, a much simpler means of describing the soil habitat would be a desirable step in both descriptive and predictive modeling of BSC distribution and floristic patterns. Biological soil crusts are easily damaged or destroyed by compressional forces associated with several disturbance types, including off road vehicles, foot traffic of humans and other animals, and mining and associated exploration activities (reviewed in Belnap & Eldridge 2003). Of these, livestock grazing is not the most severe (Belnap & Eldridge 2003), but is the most widespread of disturbance factors affecting the large majority of the western USA, and most arid rangelands of the world (Dregne & Chou 1992). Because BSCs are also notoriously slow to recover from disturbance, usually requiring decades (Belnap & Eldridge 2003), BSCs are rarely observed at their potential abundance, diversity, and functional state. Land managers increasingly require methods for inferring information about potential BSC communities to incorporate into assessment and monitoring protocols, and also to prioritize conservation or restoration efforts after overgrazing and other soil surface disturbances. An easy to use scheme for inferring community properties of BSCs is also potentially useful in traditional conservation applications (Goldstein 1999), such as identifying regions of high biological uniqueness or diversity (Myers et al. 2000). Thus, we developed a simple scheme of classifying soils with similar chemistry, texture, and sometimes appearance that was informative in predicting overall abundance of broad functional groups of BSCs (Table 1, Bowker et al. 2006b). The ability of this soil classification scheme to characterize BSC community composition and structure has not been studied previously. Our classification system does not require special expertise in either BSC taxonomy or ecology, or in soil science, but it does allow the user to infer consider information about the form and function of the BSC community. This classification is based solely on soil surface characteristics and can be derived from widely available information such as geologic maps or soil surveys, and characteristics that can be measured in the field (Schoeneberger et al. 1998). Our classification system included eight soil types that include the majority of the soil types found on the Colorado Plateau (Table 1). This soil classification explained 17-73% of the variance in 19 specific physico-chemical soil descriptors. We hypothesized that our soil classification system would also perform well in predicting BSC community charac-

teristics, and outperform elevation and annual precipitation as predictors. The former was expected because: (1) soil chemistry and texture (calcium carbonate, gypsum, and clay and silt content) influence inherent soil stability and nutrient availability, (2) shrinking and swelling clays create unstable substrates for slow growing soil surface organisms, and (3) increased pore space (determined by texture) provides an easier substrate for filamentous growth forms (e.g. cyanobacterial filaments). In contrast, precipitation ultimately determines activity time of these desiccation tolerant organisms, and elevation is a surrogate for temperature which strongly influences photosynthetic rates of BSC organisms (Lange et al. 1998). Material and Methods Sampling design and survey methods We conducted sampling over three years in the ca. 800 000-ha Grand Staircase-Escalante National Monument (Utah, USA). Our sampling design and survey methods were designed to capture a broad array of ecological gradients in our sampling strategy and are described in detail in Bowker et al. (2006b). We divided the study area into three precipitation brackets (≤ 20 cm.a–1, 20 - 30 cm.a–1, ≥ 30 cm.a–1). All sites were classified as one of eight mutually exclusive soil types (Table 1): bentonitic clay soils, calcareous sandy soils, non-calcareous sandy soils, gypsiferous soils, siliceous sandy soils (these are also non-calcareous, but are distinguished by large grain size, and siliceous cementing in parent materials), nonbentonitic fine soils, Kaiparowits-derived soils (a sandy textured parent material that is unique because it forms highly erodible badlands), and limestone-derived soils. Soils were assigned to these soil types using information in the GSENM Soil Survey (US Department of AgricultureNatural Resource Conservation Service [USDA-NRCS] 2005). All possible combinations (some did not exist in GSENM) of soil type × precipitation bracket were sampled and replicated (average n = 6, range 2 - 18) for a total of 114 sites. Only sites with relatively minor soil surface disturbance impacts were chosen for sampling, thus we did not have the luxury of a balanced random design. Rarity on the landscape accounted for low replication in a few soil type × precipitation combinations, but common combinations were well replicated. At the completion of sampling, we had 6 - 37 replicate sites of each soil type, and 19 - 49 replicate samples for each precipitation bracket. To measure crust cover and composition, we used a step point-intercept transect (modified from Evans & Love 1957) consisting of 300 points (spaced ca. 2 m apart) at each of the 114 sampling sites. At each point, ground cover measures were recorded, including mosses

- A simple classification of biological soil crust habitat on the Colorado Plateau -

3

Table 1. A dichotomous key and descriptions of soil types. 1a. Soils primarily consist of clay- and silt-sized particles. Clay content is almost always greater than 30%. Surfaces generally exhibit a roughened popcornlike texture due moderate to high shrink-swell capacity (≥ 4.5%).…………………Bentonitic fine soils 1b. Soils are either more sand-dominated, or in some cases roughly equal proportions of sand, and clay size fractions. Clay content is almost always less than 30%. Soil surfaces may be roughened due to frost-heaving of biological soil crusts, but shrink-swell capacity is generally low…………………2 2a. Soils contain at least 10% gypsum. Soil texture generally not sand-dominated, more often loamy. Soils often have visible exposures of white-gray gypsum beds……………….Gypsiferous soils. 2b. Soils contain less than 10% gypsum. Soil texture may be loamy, but is more often sand-dominated. Exposures of gypsum are lacking………..3 3a. Soils are non-calcareous. There is little or no effervescence of HCL, [0, 0-3, or occasionally 1-5% calcium carbonate (CaCO3) equivalent in USDA soil survey]. Textures are clearly sand-dominated……………………4 4a. Soils are highly sand-dominated (> 75% sand sized particles). May be weathered in place from siliceously-cemented bedrock (e.g. Navajo Sandstone) or eolian deposits of similar material ………Siliceous sandy soils 4b. Soils are sand-dominated but contain < 75% sand-sized particles. May be weathered in place (not from Navajo Sandstone), or on well-weathered alluvial or fluvial terraces, or derived from parent material poor in CaCO3………Non-calcareous soils 3b. Soils are calcareous. There is at least moderate effervescence of HCL at soil surface, usually > 5 maximum CaCO3 equivalent. Textures are variable………5 5a. Soils effervesce strongly, generally containing maximal CaCO3 equivalent > 15%. Derived from either Kaiparowits Formation (Fm.), Timpoweap Member (Mbr.) of Moenkopi Fm., Kaibab Fm., or non-gypsiferous, non-limestone members of the Carmel and Moenkopi Fms. …………………..6 6a. Soils are derived from the Kaiparowits Fm. and commonly occur in highly eroded, gray-colored badlands, soil has sandy salt-and-pepper appearance …………Kaiparowits-derived soils 6b. Soils are often light colored (e.g. gray, yellowish-gray, pale orange), and are derived from limestone (Timpoweap Mbr. of Moenkopi Fm. or Kaibab Fm. in GSENM)……..Limestone-derived soils 6c. Soils are deep orange or red grading into vermillion, and are derived from nongypsiferous, non-limestone members of the Moenkopi Fm. and Carmel Fm., often channery at surface… Non-bentonitic fine soils 5b. Soils are usually less effervescent, and less calcareous, but definitely effervescent. Soils are not derived from above listed formations, and do not generally occur in badlands. Most of these soils are sand-dominated, but include some loamy textured soils as well………….Calcareous sandy soils Bentonitic fine soils (B): Weathered in place from Tropic Fm., Blue Gate & Tununk Mbrs. of Mancos Fm., Brushy Basin Mbr. of Morrison Fm., and various shale members of the Chinle Fm.; Colors variable but often impressive, may include grays, reds, mauve, and green; Forming badland landscapes supporting very sparse vegetation, including the dwarf Atriplex corrugata; highly effervescent due to CaCO3 and other carbonates; may be somewhat gypsiferous, but not containing whitish gypsum beds; clay sized particles predominate and smectitic clay minerals generate high shrink-swell capacity; Generally lacking in BSC cover. Gypsiferous soils (G): Weathered in place from Shnabkaib Mbr. of Moenkopi Fm., Crystal Creek and Paria River Mbrs. of Carmel Fm., and Paradox Fm.; Gypsum-rich patches generally whitish or grayish, surrounding soil surface is reddish in many cases; Vegetation variable (Artemisia, Atriplex, Ephedra and Pinus-Juniperus); Containing CaCO3, but gypsum always more abundant; Contain beds or pockets of gypsum, including alabaster cliffs in some cases; Gypsum-rich microsites dominated by silt and clay sized particles, surrounding matrix may be more sandy; Potentially supporting high cover of moss and lichen rich BSCs. Siliceous sandy soils (S): Weathered in place or wind transported from non-calcareous siliceously-cemented sandstones (e.g. Navajo Sandstone and Coconino Sandstone); Color may be nearly white to pink to deep orange; Vegetation variable, depth-dependent (includes desert grasslands and Artemisia communities); CaCO3 and gypsum lacking; Texture strongly skewed toward coarse sand; Potentially supporting high cover of cyanobacterial soil crusts. Non-calcareous sandy soils (NC ): Derived from various parent materials (usually weakly calcareous, including Judd Hollow Mbr. of Straight Cliffs Fm., or Shinarump Mbr. of Chinle Fm. but exclusive of siliceous sandstones), but surface weathering has been strong enough to leach out CaCO3; Colors vary, dependent on parent material; Vegetation various, commonly Artemisia; Most abundant in relatively high precipitation areas due to greater leaching; Containing little or no CaCO3 or gypsum; Dominated by sand sized particles (often poorly sorted), but generally less coarse than Siliceous sandy soils; Potentially supporting moderate to high cover of cyanobacterially dominated BSCs, with a rich moss and lichen component. Kaiparowits-derived soils (K): Weathered in place from the Kaiparowits Fm. Occur in gray colored badlands, with variable climate-dependent vegetation; Violently effervescent due to carbonates, lacking gypsum; Texture is sand-dominated, with distinctive salt-and-pepper appearance. Generally lacking BSC cover. Limestone-derived soils (L): Weathered in place from any limestone (e.g. Kaibab Fm. or the Timpoweap Mbr. of Moenkopi Fm.; Colors range from grays to pale yellow or orange coloration (dependent on parent material); Vegetation climate-dependent, usually Artemisia or Pinus-Juniperus dominated; Moderately effervescent due to presence of CaCO3, gypsum not abundant; Texture generally loamy, but dependent on parent material; Potentially supporting high cover of moss and lichen rich BSCs. Non-bentonitic fine soils (NB): Weathered in place from non-limestone, non-gypsiferous, and non-bentonitic fluvial shale red beds (e.g. Mbrs. of Moenkopi Fm., Carmel Fm., Halgaito Fm., Organ Rock Fm., Hermit Fm.); Notably deep orange, red or purple colored; Vegetation often sparse, including Coleogyne, Artemisia, and Pinus-Juniperus; Moderately effervescent due to presence of CaCO3, gypsum not abundant; Texture usually loamy, but sand fraction is fine; conspicuous flat channers often on surface, exposed soil forms a vesicular physical crust; Potentially supporting very low to moderate cover of moss and lichen BSCs, dependent on moisture. Calcareous sandy soils (C): Derived at least partially from any calcareous sandstone, exclusive of the Kaiparowits Fm. (e.g. Kayenta Fm., Entrada Sandstone, Dakota Fm. among several others), weathering has not been sufficient to remove all CaCO3 from soil surface; Colors variable, dependent on specific parent material; Vegetation highly variable, climate dependent; Usually moderately effervescent, gypsum content minor; Texture generally skewed toward more sandy size classes, although some soils are loamy; Potentially supporting low to moderate cyanobacterial BSC cover, with some mosses and lichens.

4

Bowker, M.A. & Belnap, J

and lichens to species, light cyanobacterial BSCs (early successional BSCs, almost exclusively Microcoleus spp.), and dark cyanobacterial BSCs (later succesional BSCs dominated by Microcoleus spp., Nostoc spp. and Scytonema spp.). Statistical analysis To determine the relative contribution of three environmental predictors linked to soil stability and fertility, and physiological activity of BSC organisms (soil type, annual mean precipitation [modeled within a 5.08 cm resolution], elevation [values obtained using GPS]) to BSC richness, evenness (Pielou’s J), and community structure, we used three simple structural equation models (SEM) in Amos 5.0 (SPSS Inc., Anon. 2003). These models were essentially equivalent to multiple regression models except that they allowed the inclusion of the categorical predictor ‘soil type’ as a composite variable (Grace 2006), allowed the inclusion of correlated predictors by specifying their relationship to one another (e.g. our models specifically state that annual precipitation is influenced by elevation), and used a maximum likelihood technique to estimate regression weights simultaneously. Often, researchers using SEM advance a causal hypothesis and test the goodness of fit of their model structure. In this case, the causal ordering of the variables in the models is known with high confidence; thus, a test of fit was not of interest. Instead, because we were interested only in maximum likelihood estimates of regression weights and R2, we allowed all of the predictors to freely covary, creating a ‘saturated model’. Saturation precludes an overall goodness of fit test, which is not of consequence here. Community structure was a synthetic response variable derived from a non-metric multidimensional scaling (NMDS) ordination. We used the ‘slow and thorough’ autopilot setting in PC-ORD 4.0 (MJM Software Design 1999), which selects optimal dimensionality of the ordination using a Monte Carlo test. In this and subsequent NMDS ordinations, we used the Bray-Curtis distance measure because it is compatible with the distribution problems common to community data. Data were relativized to species maximum (i.e. data for each species was rescaled from 0-1). This transformation suppresses the effect of a few dominant taxa upon the analysis, and gives equal weight to all community members; this is valuable in broadening inference from the species to the community level. The ordination was one-dimensional, thus the axis scores were used as a univariate variable in our SEM. To determine how specific soil types differed in species richness and evenness, we used two ANOVA models in JMP IN 5.0 (SAS Inst. 2005). To determine specifically how community structure changed among

Fig. 1. Three structural equation models of identical structure using three predictors (soil type, elevation, and annual precipitation) as determinants of three responses: BSC species richness, evenness, and community structure. Rectangles represent measured variables. The composite variable (Grace 2006) ‘soil type’ is represented by a diamond. Unidirectional arrows represent a directed causal influence of one variable upon another. Bidirectional arrows represent undefined covariance between a pair of variables. Path coefficients, appearing above unidirectional arrows, are equivalent to regression weights or partial correlation coefficients (width of arrows is proportional to path coefficients). R2 appears above every endogenous variable and is interpreted as for any linear model. In all three cases, soil type is the best predictor of BSC parameters. a correlations of elevation with soil types: B = – 0.10, G = 0.26, L = 0.35, NB = 0.06, NC = 0.17, K = –0.04, S = – 0.02. b correlations of precipitation with soil types: B = – 0.24, G = – 0.11, L = 0.23, NB = 0.12, NC = 0.22, K = 0.04, S = – 0.03. **** P < 0.0001

soil types, we used indicator species analysis (Dufrene & Legendre 1997) in PC-ORD (MJM Software Design 1999). Indicator species analysis uses abundance and frequency data to yield a percent of perfect indication value for each variable-group combination and uses a Monte Carlo test to determine the probability of obtaining a given indicator value (IV) due to chance alone. We defined a ‘strong indicator’ of a particular soil type as one with an IV of 0.25-0.50, a ‘very strong indicator’ as one with an IV of 0.51-0.75, and an ‘excellent indicator’ as one with an IV ≥ 0.76.

- A simple classification of biological soil crust habitat on the Colorado Plateau -

5

Results BSC community properties as a function of environ­ mental predictors Using simple structural equation models with soil type, elevation, and precipitation as predictors, we were able to explain more than half of the variance in BSC species richness (R2 = 0.57), about a third of the variance in community structure (R2 = 0.34), and over half in BSC evenness (R2 = 0.59) (Fig. 1). In the case of richness and community structure, soil type was much more informative than the other variables, although they did make contributions. We found that in the case of evenness, soil type (r = 0.77) accounted for virtually all of the variance explained (Fig. 1). BSC community properties as a function of soil type Both species richness (F = 17.7429, P < 0.0001) and evenness (F = 68.1449, P < 0.0001) were strongly influenced by soil type (Fig. 2). Richness was about six times higher in gypsiferous soils compared to bentonitic fine soils and Kaiparowits-derived soils; the other soil types were intermediate. Evenness exhibited a similar pattern except that limestone-derived soils were the most even and were similar to gypsiferous soils; again bentonitic fine soils and Kaiparowits-derived soils were much lower (about five-fold) and other soil types were intermediate. Because BSC community structure was also largely a function of soil type, we conducted an indicator species analysis. A total of 19 out of 59 species and other taxonomic groups were strong to excellent indicators (defined as IV > 25) of individual soil types (Table 1). Bentonitic fine soils, calcareous sandy soils, Kaiparowitsderived soils and non-bentonitic fine soils had no strong indicator species. Gypsiferous soils had eight strong to excellent indicators; seven of these were very strong, (Lecanora gypsicola, Psora decipiens, Didymodon nevadensis, Diploschistes diacapsis, Squamarina lentigera, Acarospora nodulosa ssp. nodulosa), and two were excellent indicators (Catapyrenium spec., Fulgensia bracteata) primarily due to their fidelity to this soil type. Limestone-derived soils were characterized by five strong indicators, of which Aspicilia aspera and Psora cerebriformis were very strong. Siliceous sandy soils and non-calcareous sandy soils each had two strong indicators. However they were not as strong as those for the gypsiferous soils or limestone-derived soils due to lack of fidelity in the case of siliceous sandy soils, and a lack of consistency in the case of non-calcareous sandy soils.

Fig. 2. Species richness and evenness as a function of soil type. Error bars represent one standard error. B = bentonitic fine soils, K = Kaiparowits-derived soils, C = calcareous sandy soils, S = siliceous sandy soils, NB = non-bentontitic fine soils, NC = non calcareous sandy soils, G = gypsiferous soils, L = limestone-derived soils. Shared letters indicate Tukey-Kramer test, P > 0.05.

Discussion These data suggest that soil characters are by far the most influential natural abiotic predictor of the richness, evenness, and community structure of BSCs of the Colorado Plateau. A simple system of eight soil types provides a relatively non-technical means of summarizing BSC community properties. This information is potentially applicable in inferring reference states in range management (Bowker et al. 2006b) and ecological restoration (Bowker 2007), conservation of BSC function (Bowker et al. in press), and reserve design for conserving biodiversity of BSCs. A useful framework for classification of BSC habitat on the Colorado Plateau The majority of soils of the Colorado Plateau are characterized primarily by residuum weathered in place from sedimentary rocks. Of the five factors of soil formation (Jenny 1941), parent material (i.e. geology) has

6

Bowker, M.A. & Belnap, J

Table 2. Results of indicator species analysis, by community member or community type. Soil type refers to the soil type for which a community member has the highest indicator value IV. B = bentonitic fine soils, C = calcareous sandy soils, G = gypsiferous soils, K = Kaiparowits-derived soils, NB = non-bentontitic fine soils, NC = non calcareous sandy soils, S = siliceous sandy soils. Bold I.V. and associated P-values indicate species that are strong to excellent indicators. Species

Taxonomic Group

light cyanobacterial crust dark cyanobacterial crust

Cyanobacterial community Cyanobacterial community

Aloina bifrons Aspicilia hispida Aspicilia aspera Aspicilia desertorum Aspicilia or Lecanora spec. Acarospora nodulosa Acarospora schleicheri Bryum argenteum and lanatum Bryum kunzei Bryum caespiticium Crossideum spec. Collema coccophorum Catapyrenium spec. Placidium spp. Ceratodon purpureus Cladonia pyxidata Collema tenax Candellariella terrigena Calopolaca tominii Caloplaca jungermanii Caloplaca lactea Diploschistes muscorum Diploschistes diacapsis Didymodon nevadensis Desmatodon spec. Didymodon vinealis Endocarpon pusillum Encalypta spp. Fulgensia desertorum Fulgensia bracteata Gypsoplaca macrophylla Grimmia orbicularis Grimmia anodon Brachythecium collinum Heppia spec. Lecanora gypsicola Leptogium lichenoides Lecanora muralis Lecanora cf. zosterae Nostoc cf. flagelliforme Rinodina spec. Psora cerebriformis Psora globifera Psora pruinosa Psora decipiens Pterygoneurum ovatum Peltigera rufescens Pterygoneurum subsessile Psora tuckermanii Peltula patellata Unknown pyrenolichen Squamarina lentigera Syntrichia caninervis Syntrichia ruralis Tortula brevipes Tortula mucronifolia Toninia sedifolia

Moss Lichen Lichen Lichen Lichen Lichen Lichen Moss Moss Moss Moss Lichen Lichen Lichen Moss Lichen Lichen Lichen Lichen Lichen Lichen Lichen Lichen Moss Moss Moss Lichen Moss Lichen Lichen Lichen Moss Moss Moss Lichen Lichen Lichen Lichen Lichen Cyanobacterium Lichen Lichen Lichen Lichen Lichen Moss Lichen Moss Lichen Lichen Lichen Lichen Moss Moss Moss Moss Lichen

Soil type

I.V.

P

C S

15.9 33.8

0.11 0.0002

G L L NC S G NC S G L G L G G NC S G NC G L L NC G G C C NB S G G G G L C NC G B NC B L NC L L NC G L NC L L G C G L S G K L

20.3 73.7 14.5 34.7 12.7 58 25 32.3 7.4 16.2 18 41.1 90 22.6 24.6 19.2 20.9 26.9 23.6 18.6 20 19 64.3 66.7 2.7 2.7 20.1 10.8 6.5 87.8 23.4 33.3 10 2.7 6.2 74.6 3.4 21.3 3.5 23.5 6.2 57.7 6.6 19.8 67.8 24.4 12.5 8.2 26.9 14.3 2.7 58.3 29.4 24.1 10.1 16.7 45.9

0.03 0.0002 0.09 0.0028 0.19 0.0002 0.010 0.003 0.57 0.09 0.03 0.0004 0.0002 0.04 0.03 0.05 0.14 0.05 0.03 0.02 0.01 0.03 0.0002 0.0002 0.06 0.13 0.91 0.0002 0.01 0.0006 0.23 0.68 0.0002 0.81 0.02 0.93 0.02 0.67 0.0002 0.48 0.04 0.0002 0.01 0.11 0.24 0.0128 0.24 0.0002 0.004 0.06 0.19 0.05 0.002

- A simple classification of biological soil crust habitat on the Colorado Plateau the dominant influence because of the very different textures and chemistries of the various rock formations and due to the relative young age of many of the soils. This differs considerably from the Great Basin, Mojave, and Sonoran Deserts where ‘older’ soils are composed primarily of transported material of various origins, and where climate, age of alluvial parent materials, and slope may play more important roles in soil development (Scull et al. 2004). In contrast, the complex and diverse mosaic of geological formations in the Colorado Plateau results in more stark differences and heterogeneity among soils, but also more predictable soil properties. Thus, considerable information can be inferred about BSCs and other biota from geological surveys. The system of eight soil functional types used here is based upon observations and formal surveys of hundreds of sites across the Colorado Plateau. Classification rules (Table 1) are based upon available information in soil surveys (e.g. soil physico-chemical properties and often parent material; USDA-NRCS, Anon. 2005), geological surveys (e.g. descriptions of parent materials; Doelling & Davis 1989), and field-observable characteristics (Schoeneberger et al.1998). Bowker et al. (2006b) found that this system explained variance in 19 different soil properties, reducing the coefficient of variation by 35% on average. It is broad enough that it can be applied to most soils of the Colorado Plateau and potentially to other regions. To truly broaden our system to cover the entire Colorado Plateau, additional soil types may need to be invoked. Basalt-derived soils, and granite-derived soils, are two potential additional groups that could be added to this classification system in the future. As with many groups of biota, BSCs appear to contain species that are generalists and specialists. At one extreme, some of the most common taxa occur in virtually all deserts of the world, e.g. Microcoleus vaginatus, Collema tenax or coccophorum, and Anomobryum argenteum (Belnap & Lange 2003). Other taxa have a high fidelity for particular soil chemical or textural characters, e.g. Aspicilia hispida or A. aspera and CaCO3, and Lecanora gypsicola and gypsum. Both calcium carbonate and gypsum, can reduce the availability of immobile nutrients such as P, and free Ca+ ions can also displace exchangeable cations such as K+ and Mg+ in cells (Lajtha & Schlesinger 1988; Bates & Farmer 1990). Habitat specialization based upon calcium compounds in the soil may reflect a disparity in mechanisms for nutrient uptake. Furthermore, in the case of gypsum specialization, sulfur may play an important role as three of our very strong gypsum indicators (Table 1; Diploschistes diacapsis, Psora decipiens, and Squamarina lentigera) are also known to inhabit the edges of sulfur-rich thermal springs (Rosentreter & Belnap 2003). Soil texture can potentially have

7

many effects. High silt and clay content is likely to increase water holding capacity of the soil which could favor less drought-resistant BSC taxa (Anderson et al. 1982). Sandy soils also tend to be less aggregated and therefore less stable, so perhaps the fastest-growing BSC organisms (the cyanobacteria) are favored on the sandier substrates (Dougill & Thomas 2004). In addition, the greater pore spaces may present fewer physical barriers to filamentous cyanobacteria. Potentially, some combination of the latter two mechanisms may have led to establishment of well-developed dark cyanobacterial crusts on siliceous sandy soils (Table 1), the sandiest soils in our study region. No species favored bentonitic clay soils, in fact these substrates were nearly devoid of BSCs (also observed in Bowker et al. 2005). Although we cannot rule out a chemical mechanism to explain this, the simplest explanation is that shrinking and swelling clays and high susceptibility to water erosion make these substrates very difficult for BSCs to colonize. In total, over half of the species encountered exhibited at least some degree of habitat specialization, and over one third of these were quite strong (Table 1). Although the mechanisms behind these physico-chemical soil preferences of specialist species are not well understood, the phenomenon is highly predictable using the soil types outlined here. The specific classification rules listed in Table 1 are designed for use across the 337 000-km2 Colorado Plateau region of the USA, but could be modified for use in other arid regions of the world. The key in accomplishing this is identification of the important abiotic gradients which predict BSC community properties, and classification of soils based upon possible combinations of these properties. Examples of the same abiotic gradients having predictive power for BSC community properties in other regions include: soil texture – Australia (Eldridge & Tozer 1997; Thompson et al. 2006), shrink-swell clays – Italy (Loppi et al. 2004), gypsum content – Tunisia and Australia (Ullman & Büdel 2003), Spain (Guerra et al. 1995), Namibia (Lalley & Viles 2005), and calcium carbonate content – Tunisia and Australia (Ullman & Büdel 2003); Columbia Basin, USA (Ponzetti & McCune 2001). Fortunately many of these characteristics are measured in or can be inferred from geological surveys and soil surveys, and in many cases maps already exist which can be used in GIS applications.

8

Bowker, M.A. & Belnap, J

An application: reserve design for conserving biodiversity of BSCs We previously developed an approach to conservation and restoration prioritization based upon BSC functional significance which uses the same predictors to determine where on the landscape BSCs may play disproportionate roles in ecosystem function, and overlays the information upon additional information describing probability of surface disturbance (Bowker et al. in press). Thus, we can identify the areas where BSCs have the greatest significance and are simultaneously most likely to be disturbed in the present or future; the former might be considered restoration priority areas, and the latter may be considered conservation priority areas. Additionally, our soil classification system has been used to model potential cover of key BSC types for use as a reference condition parameter in either range management or restoration-rehabilitation applications (Bowker et al. 2006). In contrast to these applications, when BSC biodiversity is the object of conservation, a reserve design incorporating key gradients and regional ‘hotspots’ could be more useful. Of course, by acting to conserve both organisms and function we could most comprehensively protect these systems. Traditional approaches to conservation of various groups of taxa have often sought to determine which spatial locations or habitat types support the greatest biodiversity or biological uniqueness (Myers et al. 2000, Stohlgren et al. 2005). Based upon the present work, we can clearly say that conservation of BSC taxa would be most effective via reserves on three soil types: noncalcareous sandy soils, gypsiferous soils, and limestonederived soils. These three soil types were the richest and also most even communities. Among the three groups, there were a total of 15 indicator species, indicating that these soil types contain many habitat specialists. Non-specialist species are also well represented in this soil type, indeed only a handful were never represented in either gypsiferous soils, non-calcareous sandy soils, and limestone derived soils. It appears that to a large degree, the BSC flora of other soil types are a subsets of the flora found on these three soil types. For at least three decades, gypsiferous soils have been known to harbor both a diverse and unique community of soil crusts in both North America and Europe (Anderson & Rushforth 1976; Guerra et al. 1995; Martinez et al. 2006). While fairly common in the Mediterranean regions of Europe (Martinez et al. 2006), gypsiferous soils are rare on the Colorado Plateau. Examples of this unique and poorly studied flora of Colorado Plateau gypsiferous soils include three globally rare species first described in the 1990s (Didymodon nevadensis, Lecanora gypsicola), one

more broadly distributed species that was undetected in North America until the 1980s (Acarospora nodulosa ssp. nodulosa), and one globally rare genus and species (Gypsoplaca macrophylla; St. Clair & Warrick 1987; Zander et al.1995; Rajvanshi et al. 1998). In contrast, most of the species of non-calcareous and limestone-derived soils are widespread, however the species turnover between these two habitat types is large. In addition to representing these regional-scale ‘hotspots’ that are rich in habitat specialists and are biologically unique, it is important to capture important ecological gradients in reserve design in heterogenous areas (Diamond 1975). Climatic gradients are important, but as we have shown here, edaphic gradients are more important to the structure of BSC communities. Thus, reserve design could primarily be based upon edaphic gradients. Aside from gypsum content, perhaps the most important gradient to capture is CaCO3 content. Calcareousness of substrates, either rock or soil, have long been known to influence the potential lichen and moss communities (Downing & Selkirk 1993; Ponzetti & McCune 2000). Additionally, soil texture has long been recognized as an important determinant of BSC development (Anderson et al. 1982). Thompson et al. (2006) suggest that for BSCs, small reserves would be adequate because species-area curves saturate quickly. Although we did not study species-area relationships specifically, our observations agree with this statement as long as the system of small reserves (each of perhaps several hectares) capture variance in the key gradients mentioned above. However, it is important to explicitly state that this assertion applies only to biodiversity conservation of BSC, and that small reserves cannot appreciably conserve BSC function in the larger ecosystem. Reserves are merely islands in a matrix of landscape being used for livestock production, a major negative influence upon BSC development. Incorporation of BSCs into land management applications such as reserve design and resource management plans (Bowker et al. 2006b; Bowker et al. 2007) requires the delivery of accessible tools to managers, who cannot be expected to also be BSC experts. With the application of the BSC habitat classification system described here, non-specialists can make informed decisions about where BSC biodiversity or function might be conserved, and where BSCs might be most useful as ecological indicators.

- A simple classification of biological soil crust habitat on the Colorado Plateau Acknowledgements. We thank GSENM (US Bureau of Land Management) for funding this research. Dr. N. Johnson, S. Reed, N. DeCrappeo, B. Chaudhary, A. Antoninka and M. Lau provided useful comments on early drafts. Dr. Jan Bakker and two anonymous reviewers helped us improve an early draft. We thank Drs. K. Sutcliffe and M. Miller for providing valuable consultation concerning the planning of this research. Dr. T. O’Dell, S. Stewart, H. Barber, P. Chapman, and members of the GSENM staff assisted with logistical support. Drs. J. Spence and T. Graham provided helicopter access to remote sites. L. Pfenninger, S. Bartlett, J. Brundage, K. Kurtz, C. Nelms, E. Kneller, W. Fertig, and L. Fertig provided indispensable field assistance. Drs. R. Rosentreter, L. Stark, and J. Spence provided consultation on difficult lichen and moss identifications. The use of trade, product, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government. References Anon. 2005. Grand Staircase-Escalante National Monument soil survey. USDA-NRCS, US Department of Agriculture, Washington DC, US. Anderson, D.C., Harper, K.T. & Holmgren, R.C. 1982. Factors influencing development of cryptogamic soil crusts in Utah deserts. Journal of Range Management 35: 180-185. Bates, J.W. & Farmer, A.M. 1990. An experimental study of calcium acquisition and its effects on the calcifuge moss Pleurozium schreberi. Annals of Botany (London) 65: 87-96. Belnap, J. 2002. Nitrogen fixation in biological soil crusts from southeast Utah, USA. Biology and fertility of soils 35: 128-135. Belnap, J. & Eldridge, D.J. 2003. Disturbance and recovery of biological soil crusts. In: Belnap, J. & Lange, O.L. (eds.) Biological soil crusts: Structure, function, and management, pp. 363-383, Springer-Verlag, Berlin, DE. Belnap, J. & Lange, O.L. 2003. Biological soil crusts: structure, function, and management. Springer-Verlag, Berlin, DE. Beymer, R.J. & Klopatek, J.M. 1991. Potential contribution of carbon by microphytic crusts in pinyon-juniper woodlands. Arid Soil Research and Rehabilitation 5: 187-198. Bowker, M.A., Belnap, J., Davidson, D.W. & Phillips, S.L. 2005. Evidence for micronutrient limitation of biological soil crusts: potential to impact aridlands restoration. Ecological Applications 15: 1941-1951. Bowker, M.A., Belnap, J., Davidson, D.W. & Goldstein, H. 2006a. Correlates of biological soil crust distribution across a continuum of spatial scales: support for a hierarchical conceptual model. Journal of Applied Ecology 43: 152-163. Bowker, M.A., Belnap, J. & Miller, M.E. 2006b. Spatial modeling of biological soil crusts to support rangeland assessment and monitoring. Rangeland Ecology and Management 59: 519-529. Bowker, M.A. 2007. Biological soil crust rehabilitation in theory and practice: an underexploited opportunity. Restoration Ecology 15: 13-23.

9

Bowker, M.A., Miller, M.E., Belnap, J., Johnson, N.C. & Sisk, T.D. In press. Prioritizing conservation effort through the use of biological soil crusts as ecosystem function indicators in an arid region. Conservation Biology. Brotherson, J.D., Rushforth, S.R. & Johansen, J.R. 1983. Effects of long-term grazing on cryptogam crust cover in Navajo National Monument, Arizona. Journal of Rangeland Management 36: 579-581. Diamond, J.A. 1975. The island dilemma: lessons of modern biogeographic studies on the design of natural reserves. Biological Conservation 7: 129-146. Doelling, H.H. & Davis, F.D. 1989. The geology of Kane County, Utah. Utah Department of Natural Resources Bulletin 124. Dougill, A.J. & Thomas, A.D. 2004. Kalahari sand soils: spatial heterogeneity, biological soil crusts and land degradation. Land Degradation & Development 15: 233-242. Downing, A.J. & Selkirk, P.M. 1993. Bryophytes on calcareous soils of Mungo National Park, an arid area of southern central Australia. Great Basin Naturalist 53: 13-23. Dregne, H.E., & Chou, N.-T. 1992. Global desertification dimension and costs. In: Dregne, H.E. (ed.) Degradation and restoration of arid lands, pp. 249-281. Texas Tech University, Lubbock, TX, US. Dufrene, M. & Legendre, P. 1997. Species assemblages and indicator species: the need for a flexible asymmetric approach. Ecological Monographs 67: 345-366. Eldridge, D.J. & Tozer, M.E. 1997. Environmental factors relating to the distribution of terricolous bryophytes and lichens in semi-arid eastern Australia. Bryologist 100: 28-39. Esslinger, T.L. 2008. A cumulative checklist for the lichenforming, lichenicolous and allied fungi of the continental United States and Canada. http://www.ndsu.nodak.edu/ instruct/esslinge/chcklst/chcklst7.htm [North Dakota State University]. Evans, R.A. & Love, R.M. 1957. The step-point method of sampling: a practical tool in range research. Journal of Range Management 10: 208-212. Grace, J.B. 2006. Structural equation modeling and natural systems. Cambridge University Press, Cambridge, UK. Guerra, J., Ros, R.M., Cano, M.J. & Casares, M. 1995. Gypsiferous outcrops in SE Spain, refuges of rare, vulnerable, and endangered bryophytes and lichens. Cryptogamie, Bryologie Lichenologie 16: 125-135. Jenny, H. 1941. Factors of soil formation: a system of quantititive pedology. McGraw-Hill, New York, NY, US. Jones, C.G., Lawton, J.H. & Shachak, M. 1997. Positive and negative effects of organisms as physical ecosystem engineers. Ecology 78: 1946-1957. Lajtha, K. & Schlesinger, W.H. 1988. The effect of CaCO3 on the uptake of phosphorus by two desert shrub species Larrea tridentata (D.C.) Cov. and Parthenium incanum H.B.K. Botanical Gazette 149: 328-34. Lalley, J.S. & Viles, H.A. 2005. Terricolous lichens in the northern Namib Desert of Namibia: distribution and community composition. Lichenologist 37: 77-91. Lange, O.L., Belnap, J., & Reichenberger, H. 1998. Photosynthesis of the cyanobacterial soil crust lichen Collema tenax for arid lands in southern Utah, USA: role of water con-

10

Bowker, M.A. & Belnap, J

tent on light and temperature response of CO2 exchange. Functional Ecology 12: 195-202. Loppi, S., Boscagli, A. & DeDominicus,V. 2004. Ecology of soil lichens from Pliocene clay badlands of central Italy in relation to geomorphology and vascular vegetation. Catena 55: 1-15. Martínez, I.,  Escudero, A., Maestre, F.T., de la Cruz, A., Guerrero, C. & Rubio, A. 2006. Small-scale patterns of abundance of mosses and lichens forming biological soil crusts in two semi-arid gypsum environments. Australian Journal of Botany 54: 339-348. Mazor, G., Kidron, G.J., Vanshak, A. & Abeliovich, A. 1996. The role of cyanobacterial exopolysaccharides in structuring desert microbial crusts. FEMS Microbiology Ecology 21: 121-130. Myers, N., Mittermeier, C.G., Mittermeier, G.A., da Fonseca, B. & Kent, J. 2000. Biodiversity hotspots for conservation priorities. Nature 403: 853-858. Nash III, T.H. & Moser, T.J. 1982. Vegetational and physiological patterns of lichens in North American deserts. Journal of the Hattori Botanical Laboratory 53: 331-336. Ponzetti, J.M. & McCune, B.P. 2001. Biotic soil crusts of Oregon’s shrub steppe: community composition in relation to soil chemistry, climate, and livestock activity. Bryologist 104: 212-225. Rajvanshi, V., St. Clair, L.L., Webb, B.L. & Newberry, C.C. 1998. The terricolous lichen flora of the San Rafael Swell, Emery County, UT, USA. In: Glenn, M.G., Harris, R.C., Dirig, R., Dole, M.S. (eds.) Lichenographia Thomsoniana: North American lichenology in honor of John W. Thompson., pp. 399-406. Mycotaxon, Ithaca, NY, US. Reynolds, R.L., Belnap, J., Reheis, M., Lamothe, P. & Luiszer, F. 2001. Aeolian dust in Colorado Plateau soils: nutrient

inputs and recent change in source. Proceedings of the National Academy of Science U.S.A. 98: 7123-7127. Rogers, R.W. 1972. Soil surface lichens in arid and sub-arid south-eastern Australia. III. The relationship between distribution and environment. Australian Journal of Botany 20: 301-316. Rosentreter, R. 1986. Compositional patterns within a rabbitbrush (Chrysothamnus) community of the Idaho Snake River Plain. In: McArthur, E.D., Welch, B.L. (eds.) Symposium on the biology of Artemisia and Chrysothamnus. USDA Forest Service, Ogden, UT, US. Rosentreter, R. & Belnap, J. 2003. Biological soil crusts of North America. In: Belnap, J. & Lange, O.L. (eds.) Biological soil crusts: structure, function, and management, pp. 31-50. Springer-Verlag, Berlin, DE. Scheonberger, P.J., Wysocki, D.A., Benham, E.C. & Broderson, W.D. 1998. Field book for describing and sampling soils. USDA-NRCS, Lincoln, NE, US. Scull, P., Franklin, J. & Chadwick, O.A. 2004. The application of classification tree analysis to soil type prediction in a desert landscape. Ecological Modelling 181: 1-15. St. Clair, L.L. & Warrick, R.B. 1987. Acarospora nodulosa (Duf.) Hue. var. nodulosa: a new record for North America. The Bryologist 90: 48-49. Stohlgren, T.J., Guenther, D.A., Evangelista, P.H. & Alley, N. 2005. Patterns of plant species richness, rarity, endemism, and uniqueness in an arid landscape. Ecological Applications 15: 715-725. Thompson, W.A., Eldridge, D.J. & Bonser, S.R. 2006. Structure of biological soil crust communities in Callitris glaucophylla woodlands of New South Wales, Australia. Journal of Vegetation Science 17; 271-280. Tongway, D.J. & Hindley, N. 1995. Manual for assessment of soil condition of tropical grasslands. CSIRO, Canberra, AU. Ulmann, I. & Büdel, B. 2003. Ecological determinants of species composition of biological soil crusts on a landscape scale. In: Belnap, J. & Lange, O.L. (eds.) Biological soil crusts: structure, function, and management, pp. 203-216. Springer-Verlag, Berlin, DE. Zander, R.H., Stark, L.R. & Marrs-Smith, G.1995. Didymodon nevadensis, a new species for North America, with comments on phenology. The Bryologist 98: 590-595. Zander, R.H. 2008. Bryophyte flora of North America. URL: http://www.mobot.org/plantscience/bfna/bfnamenu.htm [Missouri Botanical Garden]. Received 6 September 2007; Accepted 18 December 2007; Co-ordinating Editor: J.P. Bakker.