Western North American Naturalist

Volume 63 Number 3 Article 9

8-6-2003

Dispersal of sagebrush-steppe by the western harvester ( occidentalis)

John F. Mull Weber State University, Ogden, Utah

Follow this and additional works at: https://scholarsarchive.byu.edu/wnan

Recommended Citation Mull, John F. (2003) "Dispersal of sagebrush-steppe seeds by the western (Pogonomyrmex occidentalis)," Western North American Naturalist: Vol. 63 : No. 3 , Article 9. Available at: https://scholarsarchive.byu.edu/wnan/vol63/iss3/9

This Article is brought to you for free and open access by the Western North American Naturalist Publications at BYU ScholarsArchive. It has been accepted for inclusion in Western North American Naturalist by an authorized editor of BYU ScholarsArchive. For more information, please contact [email protected], [email protected]. Western North American Katuralist 63(3), ©2003, pp. 358-362

DISPERSAL OF SAGEBRUSH-STEPPE SEEDS BY THE WESTERN HARVESTER ANT (POGONOMYRMEX OCCIDENTALIS)

John F. ~1u1P

ABSTRACT.-Like many -harvesting , the western harvester ant (Pogonomynnex occidentalis) can act as both a seed predator and a seed disperser. Dispersal results when seeds are dropped en route to the nest, are left in nest gra. naries when colonies die or are abandoned, or are removed from granaries and discarded. in nest middens. This study e.xamined the density and species identity of seeds discarded in harvester ant nest middens and compared them \\ith those found in nearby soils. Nineteen species ofseeds were recovered from middens, compared \"ith 13 species in S.m reference areas and 9 species in adjacent disk areas. Total density ofsound seeds was nearly 3 times higher in middens than at 5 m from the nest and nearly 50 times higher than in disk soils. Moreover, 4 ofthe 6 most common species overall were significantly more abundant in middens. One species, Munro globemalJow (Sphaeralcea l1Iunroana), was recovered from nearly 50% of middens but was not found in the other 2 areas. These findings suggest that the western harvester ant is a potentially important disperser ofsome sagebmsh-steppe species.

Key words: , harvester ants, Pogonomyrmex, alll·,Jlont interaction, shrub-steppe, granivonJ.

By definition granivores are predators that phytic fungus were more likely to be discarded consume seeds and cause the death ofindivid­ than uninfected seeds. Intrinsic seed attributes, ual (Brown and Ojeda 1987); however, such as nutritional quality (Kelrick et al. 1986) they commonly lose or reject seeds that they and seed coat thickness (Hissing and Wheeler have collected and cache others that are never 1976), may also influence a seed's chances of recovered. Thus, granivores may be both preda­ being discarded. Ultimately, individuals of some tors and dispersers of a pmticular plant species species dispersed in this manner may be more (Janzen 1971, Archer and Pyke 1991, Byrne abundant near nests and have increased repro­ and Levey 1993). Even when granivores are ductive output relative to individuals growing primarily seed predators, their rare dispersal in surrounding areas (Janzen 1971, Rissing services may be highly significant from a 1986). plant's perspective (Levey and Byrne 1993). This study quantified the extent to which the Harvester ants (Pogonomyrmex spp.) are western harvester ant (Pogonomyrmex occiden­ talis) discards intact the seeds of shrub-steppe specialist seed predators that can also disperse plants upon which it forages. Specifically, the seeds in several ways (MacMallOn et al. 2000). study was concerned with determining den­ Seeds that foragers collect (at distances ranging sity, species identity, and physical attributes of from < 1 m to >20 m) and return to the nest seeds found in colony nest middens that are may eventually germinate and grow ifthe colony maintained near nest entrances of P. occiden­ dies or emigrates. More typically, dispersal taUs colonies. occurs when foragers drop seeds en route to the nest or when nest workers discard seeds STUDY SITE from nest stores into middens (relUse piles) outside the nest (Hissing 1981, Kelrick et al. The study site is located in the sagebrush­ 1986, Wolff and Debussche 1999, Detrain and steppe 8 km southwest ofKemmerer in Lincoln Tasse 2000). A variety of seed attributes may County, Wyoming, on lands of the Pittsburg influence the probability that a seed, once har­ and ~1idway Coal Mining Company. Climate vested, will be discarded. For example, Knoch is cold and arid, with mean monthly tempera­ et al. (1993) found that tall fescue (Fesfuca tures ranging from -SOC (January) to l7°C anmdinacea) seeds infected with an endo- (July). '.>Iean annual precipitation of99.6 cm is

IDepartment ofZooIog}'. 2505 Unin~rsi~' Circle. Weber Stale Universil)', Ogden. UT 8-¥lO!l.

358 2003] HA..RVESTER ANTS AND SEED DiSPERSAL 359

highly variable and falls mainly as snow (Par­ (1981) "apparently viable" and to Young et a1.'s menter and MacMahon 1983). Vegetation at (1983) "potentially germinable" seed. I used the site is dominated by big sagebrush (Artemi­ only sound seeds in the analyses. Seeds were sia tridentata) and rabbitbrush (Chrysotham­ identified to species using a reference collec­ nus viscidiflorus). Other shrub species include tion previously established for the site. The winterfat (Ceratoides lanata), bitterbrush (Pur­ term "seeds" is used broadly here to refer to shia tridentata), gray horsebrush (Tetradymia the dispersal units, or diaspores, ofeach species, eaneseens), and Gardner's saltbush (Atriplex most ofwhich are actually . gardneri). Dominant understory plants are 3 Differences in total seed densities among perennial grasses, needle-anci-thread grass the 3 sampling locations were analyzed with a (Slipa eomata), Indian ricegrass (Oryzapsis i-way ANOVA. A Tukey-Kramer test was used hymenoides), and bluegrass (Poa spp.), and one to test for painvise differences (Wilkinson et annual grass, cheatgrass (Bromus tectorum). al. 1992). All other statistical tests involved Density of harvester ant colonies at the site is comparisons of individual species in the mid­ >30 . ha-] (Parmenter and MacMahon 1983). den and 5-m samples because disk samples contained too few seeds of most species to be METHODS included in the analyses of individual species. Densities of the 4 most abundant species were Soil samples were collected at 30 harvester compared using t tests. Chi-square goodness­ ant colonies in July 1994. To determine whether of-fit tests were used to compare total number density and species composition of seeds in of midden and 5-m samples containing the 6 colony middens differed from those in adja­ most common species of seeds. cent soils and in the colony foraging area, I took samples from 3 locations at each colony: a RESULTS midden located near the nest mound entrance, a point on the disk (= denuded area surround­ I recovered 19 species of seeds from mid­ den samples, 13 species from 5-rn samples and ing the mound) located 1 In from the entrance, and a point well beyond the disk located 5 m 9 from disk samples. Soil seed densities were nearly 3 times higher in middens than in the from the mound center. Thus, I collected a 5-m reference areas and nearly 50 times total of 90 samples. Disk and 5-m samples higher than in disk samples (Fig. 1). were taken in randomly chosen directions. Though seed densities in littered, undershrub locations at this site are much higher on aver­ age than those in open, intershrub spaces (Mull and MacMahon 1996), I made no attempt to 1000"

stratify the 5-m samples to account for this dif­ ~ N T ference. Thus, reported seed densities for the e ~ ~ 5-m samples are a composite of undershrub "C 7S0(} '" and interspace samples. '"~ • Samples were extracted with a cylindrical 0 sampling core having a cross-sectional area of ~= >, SOO(} 40 cm2. Samples were returned to the labora­ •• -~ tory and placed in a freezer until processing. = T Each sample was sieved to remove the fine "C'" "C 2S0(} mineral portion «0.25 mm). Seeds were then '" recovered from the remainder of the sample CJJ'" while it was viewed under a dissecting micro­ ~ 0 ~ scope. Recovered seeds were classified as either Midden Disk Sm sound or unsound. Sound seeds were firm under the pressure of forceps and were not discolored or otherwise visibly degraded; un­ Fig. 1. Mean soil seed densities in 3 sample types taken sound seeds were not firm under pressure and at 30 harvester ant colonies. Seed densities differed signif­ icantly among sampling locations (F = 31.56, P < 0.0001). were either discolored or degraded. As used All 3 pairwise comparisons were significant (P < 0.05) here, "'sound seed" is comparable to Roberts' based on a Tukey-Kramer test. Error bars are 1 Sf' 360 WESTERN NORTH AMERICAN NATURALIST [Volume 63

Four of the 6 most abundant species in soil sities (Tables 1, 2) are either common in the samples occurred more frequently in middens seed pool at the site (B. tectorum, A. deserto­ than in the 5-m reference areas (Table 1). Most rum, 0. hymenoides, and C. poroiflora) or may notably, seeds of Munro globemalJow (Sphaer­ be difficult for harvester ants to process and alcea munroana) were found in nearly half of consume once they have been collected (5. the midden samples but were not present in munroana and L. redowskii). Rejection of pre­ any ofthe 5-m samples. viously collected seeds is a behavior that has Three of the 4 most abundant species were been frequently obsen1ed in harvester ants. present in significantly higher densities in Explanations proposed to account for the re­ middens (Table 2). Midden densities ofthe 4th jection ofviable seeds by harvester ants, though species, B. tectorom, were not significantly varied and sometimes speculative. include the different from those in 5-m reference areas addition of more preferred species to nest but were still much higher than those in adja­ stores, differential rates of infection \vith fun­ cent disk areas (mean = 32.5, Sx = 19.6 seeds gal endophytes, seed novelty, and "mistakes" . m-2; t = 4.27, df = 58, P < 0.001), indicating made by nest workers (Rissing 1981, Kelrick that harvester ants discarded large numbers of et al. 1986, Knoch et al. 1993,Wolff and intact B. tectorum seeds. In addition, 87% of Debussche 1999, MacMahon et al. 2000). all midden samples contained at least 1 (mean Non-seed tissues attached to seeds may no./sample = 2.5, Sx = 0.49) discarded B. tec­ lower granivore preference because they in­ tarum caryopsis from which ants had clipped crease handling time costs and contain a high the embryo. Middens at all colonies contained proportion of indigestible structural carbohy­ various quantities of partially chewed and drates and other nonnutritive matter (Kelrick fragmented B. tectorum caryopses as well. et aI. 1986). The 2 species most disproportion­ ately represented in midden samples, S. munro­ DISCUSSION ana and L. redowskii, have seed morphologies that would make it costly for nest workers to Seed banks ofarid and semiarid ecosystems process them. The nutlet ofL. redowskii is hard show considerable spatial heterogeneity across and armed with stiff marginal prickles (Shaw the soil surface, with both biotic and abiotic 1989); the seed coat of S. munroana is thick factors contributing to this heterogeneity (Price and very hard. Rissing and Wheeler (1976) and Reichman 1987, Crist and MacMahon suggested that Chorizanthae rigida, an annual 1992, Mull and MacMahon 1996). This study of the Mojave Desert, is often discarded into found that the collection and subsequent dis­ refuse piles of Messor (=Veromessor) pergan­ carding of seeds by harvester ants is an addi­ dei because of its hard seed coat. In 2 other tional source ofspatial variability in seed density. reported cases, Pogonomyrmex workers dis­ Ants create small, high-density seed patches carded the thick-seeded propagules of Opun­ associated with middens that bave higher seed tia spp. (Gonzalez-Espinosa and Quintana­ densities than either adjacent disk areas or Ascencio 1986) and Datura discolor (O'Dowd areas beyond the nest within the colony forag­ and Hay 1980) into their nest middens after ing range (Fig. 1). Densities recorded at 5 m removing the associated fleshy pulp and elaio­ from the nest in this study are comparable to some (= oil body), respectively, of these 2 those reported for various parts of colony for­ species. aging areas sampled in previous studies at the In their early review of ant foraging ecol­ site (Crist and MacMahon 1992, Mull and ogy, Carroll and Janzen (1973) speculated that MacMahon 1996). This indicates that midden seed-eating ants may be capable of finely por­ seed densities, though greatly limited in their tioning collected seeds by eating only the spatial extent, exceed those in most of the embryo and discarding other portions. ~1 id­ background seed pool. dens of most colonies contained clipped B. As is true of aridland seed bank composi­ tectorum caryopses, indicating that ants pref­ tion in general, most ofthe 19 species found in erentially consumed the embryo. Embryo and middens were rare, occurring in 15% or fewer endosperm portions of B. tectorum seeds dif­ of all samples. Those species that occurred in fer only slightly in their nutritional content at least 30% of all midden samples or that (T.O. Crist personal communication), but it is were present in midden samples at high den- unclear if such small differences are sufficient 2003] HARVESTER MTS AND SEED D,SPERSAL 361

TABLE 1. Chi-square tests comparing totaJ number of midden and 5-m soil samples (n = 30 for each type of sample) containing seeds ofthe 6 most common species. For all comparisons df = 1. Sampling location Species Midden om Chi-square P Bromu$ tectomm 24 19 0.58 > 0.25 Sphaeralcea mUl1rQOna 13 0 13.0 < 0.001 Lappula redowskii 26 8 9.52 < 0.005 Alyssum deserlomm 27 17 2.26 > 0.10 Collinsia pa!'Viflora 15 6 3.84 = 0.05 Oryzopsis hymitnoides 9 2 4.46 < 0.05

TABLE 2. Comparison ofmean seed densities in soil samples taken in midden and 5-m locations (n = 30 for each type ofsample) for the 4 most abundant species. For all comparisons df = 58. Mean (sr) no. seeds. m-2 Species Y1idden 5m i-value P Bromus tectorum 1200.3 (269.6) 707.5 (185.1) 1.51 > 0.10 Sphaeralcea munroollo 557.5 (189.2) 0 3.03 < 0.005 Lappula redowskii 1307.5 (286.1) 250.1 (119.3) 3.41 < 0.005 Alyssum desertorum 3725.6 (538.2) 1390 (325.7) 2.34 < 0.05

to cause the observed consumption pattern. For example, the strong and consistent winds Alternatively, nest workers may remove the typical of the sagebrush-steppe (Allen et al. embryo from Bromus seeds to prevent germi­ 1989) coupled with the periodic surface flow of nation and then store or feed to the larvae water (Anderson and Mull 1993) quickly remove both portions of the caryopsis. However, this most litter, including seeds, from intershrub seems unlikely given that no discarded B. tec­ spaces and deposit it beneath shrubs (Kelrick tarum embryos were found in refuse piles. 1991). Thus, detailed studies ofthe fate ofhar­ Preferential consumption ofgrass seed embryos vester ant-dispersed seeds would be required has also been observed in Tetramorium caespi­ to determine if P. occidentalis provides high­ tum, Messor structor, and M. capitatus (Brian quality as well as high-quantity dispersal. 1983). Risch and Carroll (1986) reported that Solenopsis geminata removed grass seed em­ ACKNOWLEDC.\-IENTS bryos before storing the seeds, but they did not specify whether seed embryos were eaten 1 thank the Pittsburgh and Midway Coal in preference to the non-embl)'o portions. Mining Company for allowing me access to This study has demonstrated that P. occi­ their land, Annie Mull for helping \vith sample dentalis individuals deposit many intact seeds collection, processing, and analysis, and Craig of a variety of species into their middens. Anderson and one anonymous reviewer for Because nest maintenance workers rapidly improving the manuscript. remove seedlings that emerge on the disk, middens are not reliable safe sites for dis­ LiTERATURE CiTED carded seeds. Given this feature of ant behav­ ior, the effectiveness (sensu Schupp 1993) of ALLEN, :\1. F., L.E. HIPPS, AND G.E. WOOLRIDGE. 1989. Wind dispersal and subsequent establishment of VA harvester ant seed dispersal will depend upon mycorrhizal fungi across a successional arid land· the extent to which discarded seeds are trans­ scape. Landscape Ecology 2:165-171. ported away from the disk and to sites more A'\'DERSON, c.J., AND J.F. MULL. 1993. Emigration ofa colony favorable for germination and establishment. ofPog01wmyrme:r occidentalis Cresson (: Potential vectors for such transport include Formicidae) in southwestern Wyoming. Journal of the Kansas Entomological Society 65:456-458. seed-caching rodents that may occasionally ARCHER, S., AND D.A. Pl'KE. 1991. Plant· interac­ forage in middens, as well as wind and water. tions affecting plant establishment and persistence 362 \-VESTERN NORTH AMERICA.!'f NATURALIST [Volume 63

on revegetated rangeland. Joumal of Range Manage­ ~1ULL, J.E, AND lA. tvf..\.CMAHOK. 1996. Factors determin­ ment 44:558-565. ing the spatial variability of seed densities in a BRIAN, M.Y. 1983. Social ; ecology and behavioral shmb-steppe ecosystem: the role of harvester ants. biology. Chapman and Hall, London, UK. 371 pp. Journal ofArid Environments 32:181-192. BROW:'Ii, j.H., A:-.lO R.A. OlEO,'. 1987. Granivory: patterns, 0'00\\'0, D.J., AKD M.E. HAY. 1980. between processes, and consequences of seed consumption harvester ants and a desert ephemeral: seed escape on two continents. Revista Chilena de Historia Nat· from rodents. Ecology 61:531-540. ural 60,337-349. PAR,\IE~""ER, RR., A:'l/O J->\.. ~1ACMAHON. 1983. Factors de­ BYR~E. ~lll.!llll., A~D D.}. LEVEY. 1993. Removal of seeds termining the abundance and distribution ofrodents from frugivore defecation by ants in a Costa Rican in a shmb·steppe ecosystem: the role of shrubs. rain forest. Vegetatio 107/108:363---374. Oecologia 59: 145-156. CARHOLL, C.R., AND D.H. JA~ZEN. 1973. Ecology of forag­ PRICE, ~1.V, AND O.J. REICH;>.IAN. 1987. Distribution of ing by ants. Annual Review of Ecology and System­ seeds in Sonoran Desert soils: implications for het­ atics 4:231-257. eromyid rodent foraging. Ecology 68:1797-1811. CRIST, T.O., A.'1D J.A. MAC~1..o\.HO~. 1992. Hanrester ant for­ RISCH,S.] .• AND C.R. CARROLL. 1986. Effect of seed pre­ aging and shrub-steppe seeds: interactions of seed dation by a tropical ant on competition among resources and seed use. Ecology 73:1768--1779. weeds. Ecology 67:1319-1327. DETRAI:-I, C...... '10 O. TASSE. 2000. Seed drops and caches RISSI:-Ic, S.W 1981. Foraging specializations of individual by the harvester ant Messor barbarus: Do they con­ seed-harvester ants. Behavioral Ecology and Socio­ tribute to seed dispersal in Mediterranean grass­ biology 9:14-152. lands? Naturwissenschaften 87:373--376. . 1986. Indirect effects of granivol)' by harvester GONZALEZ-EsPI~OSA. ~1 .• AND M.E QUII\TANA-ASCENCIO. --~a-nts: plant species composition and reproductive 1986. Seed and dispersal in a dominant increase near ant nests. Oecologia 68:231-234. desert plant: opuntia. ants. birds, and mammals. RISSIKG, S.W, AND J. WHEELER. 1976. Foraging responses Pages 273-284 ill T.H. Fleming, editor, Frugivores ofVeromessor pergandei to changes in seed produc­ and seed dispersal. Dr. \Y. Junk, Dordrecht, The tion. Pan-Paci6c Entomologist 52:63-72. Netherlands. ROBERTS, ILo\. 19tH. Seed banks in soil. Advances in Applied JA..'1ZEN. D.H. 1971. Seed predation by . Annual Biology 6:1-55. Review of Ecology and Systematics 2:465--492. SCHUPp, E.\Y. 1993. Quantity, quality, and the effectiveness KELRICK. M.l. 1991. Factors affecting seeds in a sage­ of seed dispersal by animals. Vegetatio 107/108: brush-steppe ecosystem and implications for the dis­ 15-29. persion of an annual plant species. cheatgrass (Bro­ SHAW, RJ. 1989. Vascular plants of northern Utah: an IIllIS lectOl'l.Im L.). Doctoral dissertation, Utah State identification manual. Utah State University Press, University. Logan. Logan. 412 pp. KELRICK, i\U.• lA. MACMAHON. R.R. PAR~IENTER, AND WILKINSON, L., ~1. HILL, AND E. V"''IG. 1992. $YSTAT sta­ D.V SISSON. 1986. Native seed preferences ofshrub­ tistics version, 5.2. edition. SYSTAT, lnc. Evanston. IL. steppe rodents. binls. and ants: the relationships of WOLfF, A., AND r-.'1. DEBUSSCHE. 1999. Ants as seed dis­ seed attributes and seed use. Oecologia 68:327-337. persers in a Mediterranean old-field succession. Oikas KNOCH. T.R.• S.H. FAETH. AND D.L. Afu~OTf. 1993. Endo­ 84:443-452. phytic fungi alter foraging and dispersal by desert YOU~c, J.A., R.A. EvA..'Is, A:-IO B.A. ROUND\'. 1983. Quan­ seed-harvesting ants. Oecologia 95:470-473. tity and germinability ofOryzopsis hymenoides seed LEVEY, D.J, AND j\·f .M. BYRNE. 1993. Complex ant-plant in Lahontan sands. Journal of Range Management interactions: rain forest ants as secondary dispersers 36$2-86. and post-dispersal seed predators. Ecology 74: 1802-1812. Received 3 April 2002 ~1ACMAHO;';. J.A .. J.E ~'IULL, A:\'O T.O. CRIST. 2000. Har­ Accepted 15July 2002 vester ants (Pogonomynller spp.): their (.'Ommunity and ecosystem influences. Annual Review of Ecol­ ogy and Systematics 31:265-291.