Mcdonald, A. K., R. J. Kinucan, and L. E. Loomis. 2009. Ecohydrological

ECOHYDROLOGY Ecohydrol. 2, 66-71 (2009) Published online 29 December 200S in Wiley InterScience (www.interscience.wiley.com) DOl: 1O.IOO2/eco.40

Ecohydrological interactions within banded vegetation in the northeastern Chihuahuan Desert, USA

Alyson K. McDonald,1 * Robert J. Kinucan2 and Lynn E. Loomis3 I Texas AgriUfe Extension Service, Department Ecosystem Science and MlJTUlgement, Texas A&M University, College Station, Texas 79735, USA 2 Departmenl of Natural Resource MlJTUlgement, Sui Ross State University, Alpine, Texas 79832, USA 3 USDA Natural Resources Conservation Service, Marfa, Texas 79843, USA

ABSTRACT Landscape patterns, consisting of alternating densely vegetated bands and sparsely vegetated interbands, occur in semi-arid and arid regions of Africa, Asia, Australia and North America. The structure of vegetation patterns has been well documented, but a wide array of underlying environmental factors and ecological processes have been suggested, with no consensus regarding the genesis and persistence of these patterns. The purpose of this study was to assess ecohydrological interactions within this banded pattern by quantifying reallocation of rainfall and soil sediments. Even subtle redistribution impacted plant biomass production and species composition. Although runoff losses from interbands accounted for only 4% total rainfall, reallocation supported tree species and bunchgrasses that would not be sustained if precipitation were evenly distributed. Additionally, a rainfall threshold was identified. When storm totals exceeded 16 mm, even the densely vegetated bands were unable to capture all the rainfall and run-on from the upslope sodgrass interbands; a portion of the runoff from interbands flowed through the vegetated bands and continued downslope into the next interband. Copyright © 2008 John Wiley & Sons, Ltd.

KEY WORDS resource reallocation; source; sink; runoff; infiltration; semi-arid

Received 28 April 2008; Accepted 16 November 2008

INTRODUCTION from 50 to 820 mm. Commonalities include gentle Closely spaced, alternating dense and sparse vegeta slope gradients «3%), runoff in the form of sheet tion form landscape patterns that are easily recog flow and surface soils that are prone to physical crust nized on aerial photos but may be undetectable in the ing. However, banded vegetation is absent in some field (Lefever and Lejeune, 1997). Vegetation may be areas that share similar features. Thus, further inquiry arranged in dense bands, stripes and stipples or spots is needed. with a matrix of sparser vegetation and bare soil (a A variety of environmental factors and processes have two-phase mosaic) (d'Herbes et aI., 2001). Some pat been implicated in pattern formation and function includ terns are comprised solely of herbaceous plants (Mac ing disturbance (Hemming, 1965; Dunkerley, 2(02), aeo Fadyen, 1950; Worrall, 1959; Hemming, 1965) but most are mixed stands of woody and herbaceous species lian forces (Clayton, 1966), topography (Sherratt, 2005), (White, 1969; Chappell et al., 1999). There are region edaphic heterogeneity (Warnock, 1997), burrowing ani specific common names for these patterns. Banded and mals and soil fauna (Whitford, 1998; Ludwig et aI., striped patterns in Africa are often referred to as 'tiger 2005), plant-induced abiotic feedbacks (Couteron and bush' (WU et aI., 2000). Mogote is the regional name Lejeune, 2001; Lejeune et al., 2004; D'Odorico et aI., used in northern Mexico (Lopez-Portillo and Mon 2006; Gilad et aI., 2007) and climate change (Barbier tana., 1999). Acacia aneura (mulga) is the dominant et al., 2006). woody component in large expanses of banded vegeta Cearley (1996) described a banded vegetation sequence tion in Australia. Hence, they are referred to simply as in the northeastern Chihuahuan Desert. He attributed the 'mulga'. A summary of the extensive body of work on banded pattern to surface resource partitioning of rainfall and sed vegetation, dating back to 1941, can be found in Tong iment, variability in soil depth, or both. Warnock (1997) way et al. (2001). These two-phase vegetation mosaics completed a detailed soil survey of the pattern and con are associated with various geologic parent material, cluded that sparse and dense vegetation coincided with soil types and plant species and have been docu shallow and deep soil depths, respectively, to a root mented in regions where annual precipitation ranges limiting layer. The objective of this field study was to quantify redistribution of rainfall runoff and soil sedi *Correspondence to: Alyson K. McDonald, Texas AgriLife Extension ments within a vegetation sequence and test the hypoth Service, Department Ecosystem Science and Management, Texas A&M esis that redistribution contributes to pattern genesis and University, Fort Stockton, Texas 79735, USA. E-mail: [email protected] persistence.

METHODS a mosaic of mesquite (Prosopis glandulosa)-sodgrass and juniper (Juniperus pinchotii)-bunchgrass cts. During Study area this investigation a fourth plant community, dominated The study site is situated on a gently sloping (1·3% by P. glandulosa and Buchloe dactyloides, was distin gradient), slightly concave pediment below Cretaceous guished from the juniper-bunchgrass ct. The delineation limestone hills (Cearley, 1996; Warnock, 1997). Surface was based on species composition and productivity. B. soil textures are silt loam and silty clay loam. The win dactyloides in the mesquite-sodgrass ct was dead or dor ter season is typically cool and dry. Spring and summer mant, but still standing in the interspaces among the trees. are warm and dry punctuated with rainfall in the form Bouteloua curtipendula was the dominant bunchgrass in of convective thunderstorms. Mean annual precipitation the juniper-bunchgrass ct. at Ft. Stockton, located 50 km NW of the study site, is 310 mm (CV = 42%) (National Climate Data Cen ter, 2005), whereas annual potential evapotranspiration Field procedures is 973 mm (USDA Risk Management Agency, 2004) Beginning in May 1998, precipitation and runoff were (Figure 1). Most rainfall (74%) occurs between May and collected measured for 1 year. Runoff behaviour was October. The coldest month is January, with a mean air evaluated for two complete pattern sequences. Five temperature of 8°C. Maximum mean air temperature is 1 m x 1 m microcatchments were established within 28°C in July. each of the four plant communities for a total of 40 micro Four distinct plant community types (cts) occurred con catchments (Figure 3). Metal frames, inserted approxi tinuously as a parallel sequence downslope (Figure 2); mately 5 cm into the soil, formed the perimeter of the sequence lengths ranged from 75 to 190 m (Cearley, microcatchments. Infiltration was assumed to be equal to 1996). Proceeding downslope, an interband consisting rainfall volume minus runoff volume. Canopy intercep of widely scattered tarbush (Flourensia cernua) with a tion, stemflow and throughfall were not considered in this sparse cover of Scleropogon brevifolius sodgrass, was study. followed by a band of tarbush and a dense cover of Runoff flowed from each microcatchment into a collec Aristida purpurea bunchgrass, which was succeeded by tion bag situated in a buried bucket. Runoff was measured to the nearest 10m!. Sediment was allowed to precipi ~ ~ tate, to get filtered and oven-dried following procedures 250 r------, 30 described in Thurow et al. (1986). Soil was excavated

25 with a bucket auger and depth of wetting front was 200 recorded adjacent to each plot after every storm event.

/,..,,.., ...... PET 20 Percent basal plant cover in each microcatchment was 150 · / , estimated in August 1998, October 1998, and March / , / , 15 1999. Plant species nomenclature followed Hatch et at. 100 / / Moisture Deficit ' , (1990). Percent cover of biological soil crusts, rock / , 10 / , fragments, plant litter and mineral soil was also estimated / , 50 / , on the same three dates. In March 1999, a ten-point / , " , frame was used to measure microrelief at three locations .; " within each microcatchment. A bucket auger was used to F M A M determine depth to root restrictive layer adjacent to each Figure I. Mean monthly precipitation (PPT), potential evapotranspiration microcatchment (McDonald, 2001). (PET) and temperature (TEMP) 1971-2000 at Ft. Stockton. Pecos County. Texas.

Figure 2. Digital aerial photograph with delineations of four plant com Figure 3. Photograph of runoff microcatchment in tarbush-sodgrass plant munities within a banded pauern in the NE Chihuahuan Desert, USA. community within a banded pattern in the NE Chlhuahuan Desert, USA.

Calculation of rainfall redistribution ,n '" ~~-

Statistical analyses ,n",,,,,,,U'" All statistical analyses were conducted using SAS Version ... Q) ~ ~ 0 ~ ~ VI

,n RESULTS soil crust cover was highest in the mesquite-sodgrass ct, which also exhibited the highest percent exposed mineral soil (> 70%, Table I). Litter cover, in the form of juniper duff, was consistently highest in the juniper-bunchgrass ro"'CO'" ",,nCO'" CO

Runoff ... Cumulative rainfall for the 12-1110nth study period was Q) >o 186 mm. Storm depth inAuenced runoff response. Small U rainfall events « 16 mm) generally initiated runoff from ~

Copyright © 2008 John Wiley & Sons, Ltd. Ecohydrol. 2, 66-71 (2009) 001: 10.1002/eco ECOHYDROLOGICAL INTERACTIONS IN CHIHUAHUAN DESERT 69 the sodgrass cts, but not from the bunchgrass communi Wetting depth ties. In all communities, runoff was less than 6% dur Mean wetting depth was greatest in the tarbush-bunch ing small storms. Percent contributions to runoff dur grass ct, followed by the juniper-bunchgrass ct (Figure 6). ing large storms were 24, 3, 20 and 8%, respectively. Infiltration depth was significantly less in the sodgrass cts. Runoff volume (ml m-2) differed markedly among the When analysed by storm class, there was no significant four cts (Figure 4). Microcatchments in the tarbush difference among cts for small storms, but large storms sodgrass ct yielded the greatest runoff followed by induced significantly more infiltration in bunchgrass cts. the mesquite-sod grass microcatchments. The tarbush sodgrass ct shed seven times more water per unit area than tarbush-bunchgrass ct and three times more than Resource redistribution juniper-bunchgrass ct. The study site encompassed about 2·5 ha. The tarbush sodgrass ct occupied 1·65 ha (65%) of the site, tarbush Sediment loss bunchgrass ct 0·53 (20%), mesquite-sodgrass ct 0·17 ha (7%) and juniper-bunchgrass ct 0·19 ha (8%). The ratio Sediment loss varied with storm size (Figure 5). During of source area (runoff-generating sodgrass cts) to sink small storms, the tarbush-sodgrass ct generated the most area (run-on receiving bunchgrass cts) was 2-4: 1. 2 sediment (0·3 g m- ), whereas the other three commu Differences among wet, dry and average years are nities yielded almost none. Compared with an average characterized by the frequency and contribution of small 2 of 0·03 g m- of sediment lost in small storms in the versus large storms. During the study period, storms mesquite-sodgrass ct; large storms produced an average were almost equally divided between small and large, 2 of 15-4 g m- • Likewise, sediment production from the but large storms accounted for 91 % rainfall total. Fre 2 juniper-bunchgrass ct increased from 0 to 11·3 g m- . quencies of small storms were 79 and 87% in 1992 and Runoff from the tarbush-bunchgrass ct contained the least 1941-2000, respectively, but represented less than half 2 amount of sediment, 2·1 g m- during large storms com of the precipitation. pared with none from small events. Microcatchment runoff data were used to compute the depth of water leaving or entering each plant ct according to sequence position. Runoff depths from 7000 ~ microcatchments were mUltiplied by relative ct area. This ~o +-----~------volume was then applied to the relative area of the ct

,,; 5000 downslope. Resource redistribution was computed for three different annual precipitation totals represented by:

~ 4000 g '5 3000 Tarbush Tarbush Mesqu~e- Juniper 2 sodgrass bunchgrass sodgrass bunchgrass ::!" 2000 29m :::!' 18 m I 33 m/\ r).\ . , ~" ." 1"1 ~ \ f ~ - I~ 1000 .... ~B_. ~ ..;.~~ ... . . ~ : r-. ',.. ; I ~.~.... {'~" ·r ...... "n'-' I • ,~... o small large small large small large smaR large l1li\811 large sm8. large smell large small large o Tarbush-sodgrass Tarbush-bunchgrnss Mesquite-sodgrass Junlper-hunchgrnss

Figure 4. Mean (±I SE) runoff volumes during small and large storms -5 from four plant communities within a banded pattern in the NE Chihuabuan Desert, USA. :[ -10 ..r: 20 ,----- C. 18 +------,------ ~ -15 +-----.::..-.-+ Cl 16 +------r------ c '" 14 J------ E = -20 +------~- S 12 1-----.., ~ ~ 10 ~ 8 -25 +------ E ~ 6 (f) -30 +------~~~------~ 2 o lnIall _rge -35 -L------~ T8Ibush-sodgrass Tarbush-bunchgrass Mesquile-sodgrass Juniper-bunchgrass Figure 6. Drawing of vegetation sequence with mean plant community Figure 5. Mean (± I SE) sediment loss during small and large storms lengths. Note: drawing not to scale. Mean (±SE) wetting depths for small from four plant communities within a banded pattern in the NE and large storms within four plant communities within a banded pattern Chihuahuan Deserrt, USA. in the NE Chihuahuan Desert, USA.

1000 overall hydrologic response of the pattern to storm depth, • dry year 0 wet year C average year then rainfall characteristics are likely a fundamental E 800 f------J mechanism of this and possibly other landscape patterns. .s During large storms, this pattern may behave similar 5 600 1- •••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• ~ 639mm :;::; to a mulga pattern in Australia (Tongway and Ludwig, a.!! 1990). The mulga grove-intergrove pattern had a downs "0 400 ~ lope sequence of three plant communities: (i) ephemeral C ...... , .- ...... ··· · 310mm O; forbs and grasses forming a treeless, sparsely covered z 200 runoff zone, (ii) a grassy transfer zone immediately above a (iii) mulga grove with a dense understory of peren o nial forbs and grasses forming a run-on zone. When Tartlush- Tartlush- Mesquite- Juniper sodgrass bunchgrass sodgrass bunchgrass small storms occurred at our West Texas study site, the tarbush-sodgrass acted as a water source for the tarbush Figure 7. Net precipitation assuming 100% infiltration of run-on com puted for 1992, a dry year (186-mm rainfall), mean rainfall J941-2000 bunchgrass sink. Likewise, the mesquite-sodgrass ct sup (310-mm rainfall) and 1998, a wet year (639-mm rainfall) in four plant plied run-on to the juniper-bunchgrass sink. However, communities within a banded pattern in the NE Chihuahuan Desert, USA. during large storms, data indicate that bunchgrass com munities may act as water transfers rather than sinks; if (1) the study period (a dry year), (2) a wet year (1992) infiltration capacity is exceeded some overland flow will and (3) the long-term mean (1941-2000) (Figure 7). pass through the bunchgrass cts. Net precipitation was consistently higher in the vege The proportion of source and sink areas that develop tated bands compared with the interbands. Net precipita may be related to rainfall. In regions with relatively low tion in the tarbush-bunchgrass ct always exceeded actual rainfall, larger source areas would be necessary to support precipitation and was greatest during dry years, followed sink areas. The desert bands described by MacFadyen by wet years and reduced to only 1150/0 actual precipi (1950) and Hemming (1965) occurred in a region of tation for an average year. The opposite is true for the Somalia that receives about 150-mm rainfall annually sodgrass source cts where net precipitation is nearly equal and the source to sink ratio was 4: I compared with to measured precipitation during an average year com 2-4 : 1 found at this site in the Chihuahuan Desert, where pared with an unusually wet or dry year. annual precipitation is 310 mm. These ratios are also dynamic. Valentin and d'Herbes (1999) compared photos from 1962 to 1995 and found the band: interband ratios DISCUSSION varied from 0·5 I to 2·33. The ratio increased from 1·13 Results of this study support the assertion that eco in 1962 to 2·3 in 1992. hydrological interactions contribute to vegetation band Because microcatchments were intentionally placed ing on the Stockton plateau in West Texas. Vegeta to avoid woody plants, the influence of woody plant tion structure on the study site was strongly related canopy on resource reallocation cannot be determined by to measured rates of runoff and soil sediment losses. this study. Additionally rainfall and runoff intensity data The entire study site received the same amount of rain rather than event totals would allow us to further define fali, but we observed that water resources were not and evaluate the impact of runoff behaviour. evenly distributed across the site. Large, contiguous areas Band migration was not part of this study, but efforts of tarbush-sodgrass provided additional water resources to measure upslope movement have been performed else to mostly continuous bands of tarbush-bunchgrass situ where (Montana et al., 200 I). New recruits of woody ated immediately downslope. Small, elongated patches of species were not observed, and dead, mature trees were juniper-bunchgrass received run-on from a proportional seen along the downslope edge of the juniper-bunchgrass area of mesquite-sodgrass. band. This could be interpreted as evidence of pattern In contrast to an earlier assessment by Warnock ( 1997), migration, but Lopez-Portillo and Montana (1999) sug soil depth to root limiting petrocalcic layer did not gested that the downslope margins of mesquite stripes vary significantly among the four cts, and mean depths in the Chihuahuan Desert, Mexico, contract while the exceeded 70 cm. Antecedent moisture content was also remainder of the pattern is stable and that migration similar among communities when the study was initiated is not essential for pattern subsistence. Specifically, as (McDonald, 200 I). the mesquite stand matures evaporative demand exceeds Runoff responses observed during the study period available water supply. Individuals along the downslope were extrapolated to a relatively wet year (1992), a edge die because they do not receive sufficient run-on. relatively dry year (1998) and to the available period Band migration or possibly expansion and contraction of record (1941-2000). Redistribution of water would in response to wet and dry years could be assessed by be more pronounced during dry and wet years than remote sensing of aerial photography. Microbiotic crusts during average years. Erratic rainfall patterns favour the were ubiquitous on soil surfaces in the sodgrass (source) bunchgrass cts, not the sod grass cts. If runoff behaviour communities. Their role in infiltration and runoff warrants observed during the study period is representative of the further investigation.

ACKNOWLEDGEMENTS Lefever R, Lejeune O. 1997. On the origin of tiger bush. Bulletin of MathemaJical Biology 59: 263-294. This study was funded by a Sui Ross State Univer Lejeune 0, Tlidi M, Lefever R. 2004. Vegetation spots and stripes: sity Research Enhancement Grant. We appreciate Jerry Dissipative structures in arid landscapes. IntemaJional Journal of Quantum Chemistry 98: 261-271. Puckett for providing access to the study area and to Lopez-Portillo J, Montana e. 1999. Spatial distribution of Prosopis Ray McDonald and Terrance Young for helpful assis glandulosa var. torreyana in vegetation stripes of the southern tance in the field and Brandi Burzynski for formatting Chihuahuan Desert. Acta Oecologia 20: 197-208. Ludwig JA, Wilcox BP, Breshears DD, Tongway DJ, lmeson Ae. 2005. the figures. Illustration by Patricia Manning. We would Vegetation patches and runoff-erosion as interacting ecohydrological like to thank John Ludwig and an anonymous reviewer processes in semiarid landscapes. Ecology 86: 288-297. for their thoughtful questions and constructive comments. MacFadyen WA. 1950. Vegetation patterns in the semi-desert plains of British Somaliland. Geographical Journal 116: 199-210. McDonald AK. 200 I. Resource partitioning influences vegetation banding on the Stockton Plateau, Pecos County, Texas. M.S. Thesis, REFERENCES Sui Ross State University, Alpine, Texas. Montana C, Seghieri J, Comet A. 2001. Vegetation dynamics: recruit Barbier N, Couteron P, Lejoly J, Deblauwe V, Lejeune O. 2006. Self ment and regeneration in two-phase mosaics. In Banded Vegetation organized vegetation patterning as a fingerprint of climate and human Pa/leming in Arid and Semiarid Environments, Tongway D, Leprun impact on semi-arid ecosystems. Journal of Ecology 94: 537-547, J-C, Seghieri J (eds). Springer: New York. DOl: 10·1 II IIj.1365-2745·2006·01 126.x. National Climate Data Center. 2005. Monthly summary data Cearley KA. 1996. Landscape factors controlling banded vegetalion for NWS and cooperative U.S. stations accessed online at, pal/ems in the Stockton Plaleau, West Texas. M.S. Thesis, Sui Ross http://www.ncdc.noaa.gov. State University, Alpine, Texas. SAS Institute, Inc. 1985. SAS User's Guide: Basics, version 5 edn. SAS Chappell A, Valentin C, Warren A, Noon P, Charlton M, d'Herbes 1M. Institute, Inc.: Cary; Ne. 1999. Testing the validity of upslope migration in banded vegetation Sherratt JA. 2005. An analysis of vegetation stripe formation in semi from southwest Niger. Catena 37: 217-229. arid landscapes. Journal of Mathematical Biology 51: 183-197, DOl: Clayton WD. 1966. Vegetation ripples near Gummi, Nigeria. Journal of 10· IOO7/s00285-005-03 19-5. Ecology 54: 415-417. Thurow TL, Blackburn WH, Taylor CA Jr. 1986. Hydrologic character Cody RP, Smith JK. 1997. Applied Stalistics and the SAS Programming istics of vegetation communities as affected by livestock grazing sys Language. Simon & Schuster: Prentice Hall; NJ. tems, Edwards Plateau. Journal ofRange Management 39: 505-509. Couteron P, Lejeune O. 2001. Periodic spotted patterns in semi-arid Tongway D, Leprun J-C, Seghieri J. 2001. Banded Vegetalion Palleming vegetation explained by a propagation-inhibition model. Journal of in Arid and Semiarid Environments. Springer: New York. Ecology 89: 616-628. Tongway OJ, Ludwig JA. 1990. Vegetation and soil patterning in the d'Herbes J-M, Valentin C, Tongway OJ, Leprun J-e. 2001. Banded semi-arid mulga lands of eastern Australia Australian Journal of vegetation patterns and related structures. In Banded Vegetation Ecology 15: 23-34. Palleming in Arid and Semiarid Environments, Tongway D, Leprun USDA Risk Management Agency. 2004. University of Nebraska, J-C, Seghieri J (eds). Springer: New York. National Agricultural Decision Support System (FCICIRMA 21 E0831) D'Odorico P, Laio F, Ridolfi L. 2006. Patterns as indicators of 228). productivity enhancement by facilitation and competition in dryland Valentin C, d'Herbes 1M. 1999. Niger tiger bush as a natural water vegetation. Journal of Geophysical Research-Biogeosciences 111: harvesting system. Catena 37: 231-256. G03010 DOl: 1O·102912006JGOOOI76, 2006. Warnock BJ. 1997. Relationships between soil properties and banded Dunkerley DL. 2002. Infiltration rates and soil moisture in a groved vegetation patterns in the Stockton Plateau, Texas. M.S. Thesis, Sui mulga community near Alice Springs, arid central Australia: Ross State University, Alpine, Texas. evidence for complex internal rainwater redistribution in a runoff White LP. 1969. Vegetation arcs in Jordan. Journal of Ecology 57: runon landscape. Journal of Arid Environments 51: 199-219, 461-464. DOl: 10·1 006/jare.200 1·0941. Whitford WG. 1998. Contribution of pits dug by goannas (Varanus Environmental Systems Research Institute, Inc. 1996. Using ArcView gould;;) to the dynamics of banded mulga landscapes in eastern GIS. Environmental Systems Research Institute, Inc.: Redlands; CA. Australia. Journal of Arid Environments 40: 453-457. Gilad E, Shachak M, Meron E. 2007. Dynamics and spatial organization Worrall GA. 1959. The Butana grass patterns. Journal ofSoil Science 10: of plant communities in water-limited systems. Theoretical Population 34-53. Biology 72: 214-230. WU XB, Thurow TL, Whisenant SG. 2000. Fragmentation and changes in Hatch SL, Kancheepuram NG, Brown LE. 1990. Checklist of the hydrologic function of tiger bush landscapes, south-west Niger. Journal Vascular Plants of Texas. Texas Agricultural Experiment Station of Ecology 88: 790-800. Miscellaneous Publication-1655: College Station, TIC Hemming CF. 1965. Vegetation arcs in Somaliland. Journal of Ecology 53: 57-67.