Seed Bank and Diaspore Morphology

ECOGRAPHY 25: 336–344, 2002

Gap colonization in the Patagonian semidesert: seed bank and diaspore morphology

Roberto J. Ferna´ndez, Rodolfo A. Golluscio, Alejandro J. Bisigato and Alberto Sorianoc

Ferna´ndez, R. J., Golluscio, R. A., Bisigato, A. J. and Soriano, A. 2002. Gap colonization in the Patagonian semidesert: seed bank and diaspore morphology. – Ecography 25: 336–344.

We report work on a cold, windy South American steppe dominated by tussock grasses and shrubs of small stature that, together, cover only half of the soil surface. Our objective was to ﬁnd out why seedlings and juveniles of these dominant species are generally absent from the bare or poorly-populated spots (gaps) that exist between established individuals. We hypothesized that matrix-forming species fail to colonize gaps because of a lack of properly-placed seeds, contained in diaspores which because of their morphology are blown away from gaps that otherwise would constitute safe sites for recruitment. We evaluated diaspore size and wing loading (weight:area ratio) for all common species in the community, collected seed-bank samples in different occasions and microsites, and performed detailed ﬁeld observations for one gap-dwelling species during several years. We found that: 1) Seeds of the dominant, matrix-forming species are uncommon in the soil bank of the center of the gaps between established vegetation. 2) Seeds of the dominant species are more abundant towards the edges of the gaps than at their center. 3) Diaspores of those species present in the seed bank of the gaps are smaller than diaspores of absent species; contrary to expectations, not all gap-dwelling species had larger diaspore wing loading than non-gap species. 4) Seeds and adult densities of the most common gap species (the annual Camissonia dentata), were correlated between them and across subsequent years. We conclude that it is not an overall shortage of seeds that precludes the dominant species from becoming established in the gaps, but rather the seeds’ uneven spatial distribution. We further argue that gaps would be suitable sites for recruitment, but large diaspore size makes seeds of the dominant species to be blown away by the strong westerly winds.

R. J. Fernańdez ( [email protected]), R. A. Golluscio, A. J. Bisigato, IFEVA- Ecologıá, Uni6. de Buenos Aires – CONICET, A6. San Martıń 4453, Buenos Aires C1417DSQ, Argentina (present address of A.J.B.: Centro Nacional Patagońico, CON- ICET, Bl6d. Brown s/n, Puerto Madryn, Chubut U9120ABT, Argentina).

Vegetation in arid and semiarid zones typically displays tence of low-cover or bare patches. This work examines some sort of ‘‘mosaic structure’’ of high- and low-cover the role of seed availability in the persistence of these patches (Aguiar and Sala 1997). Substantial progress spots for a particularly windy semiarid environment. has been made in understanding the processes deter- In a cold Patagonian semidesert of southern South mining the dynamics, and especially the persistence, of America, adult shrubs and perennial tussock grasses high-cover patches (e.g. Garner and Steinberger 1989, form a high-cover vegetation matrix that covers ca 50% Schlesinger et al. 1996, Klausmeier 1999). Much less is of the soil surface. The rest is occupied by bare or understood, however, about the causes of the persis- poorly-populated spots (gaps) of an average diameter of 35 cm (Soriano et al. 1994). The dominant matrix- forming species are three small-statured shrubs c deceased.

336 ECOGRAPHY 25:3 (2002) (Mulinum spinosum (Cav.) Pers., Senecio filaginoides Desert (Reichman 1984), and the Negev (Boeken et al. DC and Adesmia campestris (Rendle) Skottsb.) and 1998). three tussock grasses (Stipa speciosa Trin. et Rupr., S. Our main hypothesis was that Patagonian matrix- humilis Cav., Poa ligularis Nees ap Steud.). Gaps host a forming species fail to colonize gaps because of a lack number of annual and small perennial forbs but are of properly-placed seeds, which in turn is caused by generally unoccupied by seedlings and juveniles of the their diaspore morphology. In this windy environment, matrix-forming species (Table 1). diaspores of certain shape and size would be unable to For the non-dominant grass Bromus pictus (Hook), remain over the relatively smooth surface of the gaps Aguiar and Sala (1997) showed that, despite an abun- that otherwise would constitute safe sites for recruit- dant seed rain, establishment in gaps is infrequent ment. Diaspore morphology determines not only pri- because the strong prevalent west winds blow seeds mary dispersal by wind (Greene and Johnson 1986), but away from these bare-soil areas (see also Alippe and also retention at the soil surface (Chambers et al. 1991). Soriano 1978). Soil-water availability during the sum- A parameter that has been successfully used to predict mer drought is higher in the center of the gaps than diaspore movement during free fall is ‘‘wing loading’’, closer to the plants encircling them (Soriano and Sala defined as the ratio between the weight and the area of 1986). The establishment of Bromus seedlings in gaps is the diaspore (Matlack 1987). For a given wind intensity possible as long as its relatively large diaspores (ca and initial height, the distance traveled is inversely 15×3 mm) are experimentally attached to the surface related to the square root of the wing loading (Aguiar et al. 1992, 1994). Theories of population dy- (Augspurger and Franson 1987, Greene and Johnson namics would classify this case as one of recruitment 1993). We reasoned that, under strong winds like the uncoupled from disturbance (Grubb 1986), in which Patagonian ones, this weight:area ratio would not only affect primary dispersal but also later movements on populations are neither strictly seed- nor microsite-lim- the soil surface. Another parameter we chose to study ited (Crawley 1990, Eriksson and Ehrleń 1992). was diaspore size because, regardless of the rugosity of Vegetation patterning may in many cases be ac- the soil surface, the smaller the diaspore, the better its counted for by after-landing movements of seeds anchorage capability (Chambers et al. 1991). (Chambers and MacMahon 1994), also known as sec- We evaluated diaspore size and wing loading for all ondary dispersal. In Patagonia, dominant grasses and common species in the community, collected seed-bank shrubs have an abundant seed production in most years samples in different microsites, and performed detailed (Soriano 1983, Fernańdez et al. 1992). We wondered field observations in one of the gap-dwelling species whether seeds of these species – having diaspores of during several years in order to evaluate four predic- different size and shape than those of Bromus – are tions that follow from our hypothesis: 1) Seeds of most able to persist in the gaps or also accumulate against matrix species are absent from, or very poorly repre- established individuals. This kind of wind piling-up has sented in, the seed bank of the gaps. Our definition of been reported for other environments, such as western seed bank is broad, and includes all viable seeds in the Australia (Mott and McComb 1977), the Sonoran surface soil. 2) The abundance of seeds of the matrix species will be much higher close to the established Table 1. Floristic composition of the gaps between Patago- vegetation (the gap edge) than towards the center of the nian bunchgrasses. Data obtained at the end of the growing season within circular 30-cm diameter plots (January 1992; gaps. 3) Diaspore morphology of species found in the n=200). Asterisks indicate those species also found in the seed bank of the gaps will be different from that of seed bank of the gaps (cf. Table 2). species absent from it. We expected gap species to have SpeciesGrowth form1 Frequency (%) smaller diaspores, with a larger wing loading, than non-gap species. A corollary from this – taking into Camissonia dentata* Annual dicot 57 account that most gap species are of a very short Microsteris gracilis* Annual dicot 39 stature – is that gap species will be poor colonizers of Poa lanuginosa Rhizomatous grass 25 Poa ligularis* Bunchgrass 13 distant gaps, thus tending to re-colonize the same gaps (seedlings) year after year. We chose to work with the most Calceolaria Rhizomatous 11 common gap colonizer, Camissonia dentata Cav. Reiche polyrhiza* deciduous dicot (synonyms given below), to test our last prediction: 4) Polygala darwiniana Stoloniferous 10 evergreen dicot The number of both extant plants and seeds of each Adesmia lotoides Rhizomatous 8 species in a given gap will be positively correlated evergreen dicot across subsequent years. Gilia laciniata* Annual dicot 7 Cerastium ar6ense Rhizomatous 4 deciduous dicot Oxalis Rhizomatous 3 squamoso-radicosa deciduous dicot Site description Other species Perennials B2 This work was performed in central-western Patagonia, 1 See Golluscio and Sala (1993) for details. in an experimental field of the Instituto Nacional de

ECOGRAPHY 25:3 (2002) 337 Tecnolog´ıa Agropecuaria (INTA) located near R´ıo fully inspected in order to return to the fine-soil frac- Mayo (45°41%S, 70°16%W). From the phytogeographic tion any material that might contain seeds. The fine soil point of view, the vegetation is typical of the Occidental was spread, as a 1-cm thick layer, on paper towels District of the Patagonian Province (Paruelo et al. placed on plastic flat containers with drainage holes. 1991). All our observations took place in a representa- Soil in the containers was spray-watered regularly while tive 2-ha plot that had been closed to large herbivores maintained in a growth chamber at 15°C with 12 h of since 1983. According to a survey performed on rela- light per day. New emergence of seedlings commonly tively undisturbed stands (Golluscio et al. 1982), the finished after 4 weeks. Seedlings were grown until iden- dominant species are: the tussock grasses Stipa speciosa tified. We are aware that the duration and the environ- (13% of cover), Stipa humilis (8%), and Poa ligularis mental conditions chosen for incubation probably did (4%), and the shrubs Senecio filaginoides (5%), Mulinum not allow the germination of all the viable seeds in the spinosum (5%), and Adesmia campestris (3%). Other, samples; thus, ours is a minimum and biased estimation less-abundant species in the matrix are the palatable of seed density (Fenner 1985). Poa lanuginosa Poir. (2%), Bromus pictus+Bromus setifolius Presl. (adding up 1%) and Hordeum comosum Presl. (negligible cover). These 10 species will be re- Are seeds of the matrix species more abundant ferred to as the ‘‘matrix-forming’’ species. towards the edges of the gaps? Soils are Aridisols (Petrocalcic Calciorthids; Gollus- cio et al. 1982); mean annual rainfall is 136 mm, with We answered this question by comparing the abun- 70% of it falling during autumn and winter. Mean dance (density and frequency) of each species’ seeds at temperatures range between 2°C (July) and 14°C (Janu- the edge vs the center of the gaps. Implicit in this ary). The months with adequate temperatures for plant approach is the assumption that diaspores are evenly growth (November–March; Fernańdez et al. 1991) distributed during primary dispersal, and that it is have severe water deficits (Sala et al. 1989, Paruelo and secondary dispersal that determines final seed pattern Sala 1995). Surface winds are strong, with an average via seed redistribution (Chambers and MacMahon velocity of4ms−1 and gusts up to 28 m s−1 (Paruelo 1994, Aguiar and Sala 1997). This ‘‘center vs edge’’ et al. 1998). The summer months during which most sampling was done in such a way as to maximize the seeds are released (January–February) show an over- chances of finding abundant seeds of the matrix species whelming dominance of the west wind direction, with because: a) it was performed in March 1998, knowing 97% of the observations corresponding to the 2709 that 1997–98 season had been very wet and seed pro- 7.5° interval (on site bi-hourly vane data recorded with duction had been high; b) samples were taken close to a 21X datalogger, Campbell Sci., Logan, UT, USA). shrubs, thus ensuring a close proximity to the seed source of matrix species: the shrubs themselves and the dense ring of tussock grasses that encircle them (Sori- ano et al. 1994). Methods Twenty gaps were chosen located east of a randomly Are seeds of the matrix species uncommon in the selected shrub, and paired (gap center and edge) sam- soil bank of the gaps? ples were taken in them. Samples were cylindrical, 7.4 cm in diameter, and 0.3 cm deep. In the laboratory, Surface soil samples (hereafter referred to as ‘‘gap- diaspores \2.5 mm were separated by sieving, iden- only’’ samples) were collected on nine occasions, from tified to species, and counted. These samples were September 1990 until January 1994, never re-sampling maintained one month at field capacity in 18×22-cm the same gap. On each date 50 gaps were selected, containers at 5°C in darkness (stratification treatment) evenly spaced at 1 m, along a 50-m transect established followed by three months at 5/15°C night/day alternat- on a random direction. The samples, taken at the center ing temperatures in order to promote germination. As of each gap, were cylindrical, 8.5 cm in diameter and 5 seeds of large-seeded species were counted without cm in depth; they were air-dried over absorbent paper knowing their viability, no comparisons were done in within hours of gathering, and stored in plastic bags at this sampling between species. Also, these samples be- room temperature. In the laboratory, samples were ing shallower than those used for the gap-only samples incubated under the environmental conditions that, (see previous section), and the germination conditions based on existing information (e.g. Soriano 1960, also different, comparisons between these two data sets Alippe and Soriano 1978, Bisigato 1994), seemed most are not valid. Within each species, we tested the null likely to trigger germination of the majority of the hypothesis of no difference in seed abundance between viable seeds of each species – both dormant and non- the edge and the center of the gaps using two criteria: dormant ones. the Kruskall-Wallis test for seed density data, and the Samples were first freed of stones using a 2.5-mm comparison of confidence intervals for frequency data, sieve. Organic debris retained by the sieve were care- assuming binomial distribution (Sokal and Rohlf 1995).

338 ECOGRAPHY 25:3 (2002) Table 2. Seed-bank density in the gaps between bunchgrasses. Seedlings emerging when 5-cm depth soil samples (28 cm2 from the center of the gaps) were watered and maintained at 15°C with a 12/12 h light/dark regime during 4 weeks (n=50 for each date). Asterisks included to facilitate comparison with other tables and ﬁgures.

Species Seed density (dm−2)

1990 1991 19921993 1993 1994

Sep Jan Jan Jun Sep Dec Jan Apr Jan

Camissonia dentata* 1.4 22.1 27.0 0.8 0.1 1.1 14.7 5.7 6.6 Microsteris gracilis* 0.6 1.8 3.1 0.1 0.0 0.0 2.7 2.0 2.4 Calceolaria polyrhiza* 0.1 1.8 1.1 0.00.0 0.0 1.4 11.2 0.7 Poa ligularis* 0.1 1.8 0.80.4 0.3 0.6 3.5 1.6 4.1 Polemonium antarctica* 0.0 0.1 0.7 0.1 0.0 0.0 0.1 0.0 0.0 Gilia laciniata* 0.0 0.4 0.0 0.00.0 0.0 0.0 0.0 0.0 Senecio ﬁlaginoides 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Bromus pictus 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.1 Unknown 1.1 5.0 0.30.0 0.0 0.0 0.1 1.1 0.4 Total 3.4 33.2 33.1 1.6 0.4 1.7 22.6 21.6 14.4

Do species in the seed bank have a distinctive at least 30 cm in diameter, free from adult bunch- diaspore morphology? grasses. For additional adult-density observations a third, parallel 75-m transect was established in a similar The diaspore morphological features (size and wing fashion in 1994. The species originally present in these loading) of 15 species were recorded from samples permanent plots are those reported in Table 1. The harvested in January 1997. Six of these species were the plots were sampled for Camissonia density in 1992, ones commonly present in the seed bank of the center 1993 and 1994. Surface soil samples (‘‘Camissonia- of gaps; the other nine were matrix species absent from only’’ samples) were taken near the center of each of it. (The ﬁve species found as established adults in the the plots in January 1992 and January 1994 (10×10× gaps but absent from their seed bank all had rhizomes 2 cm). Both 1992 and 1994 samples were used for or stolons and were not included in these analyses; cf. seed-bank determination in the same way as described Table 1.) To estimate wing loading, we determined for the gap-only samples. However, for 1994 this was diaspore weight and projected area for each species. We done only with one-half of each sample, since the other weighed ca 10 samples per species, of several diaspores half was subjected to soil texture analysis (hydrometer each, on a scale with a precision of 0.1 mg. We mea- method). sured projected area, length and width of the diaspores The spatial patterning of data in Fig. 2 was assessed by using a pseudo-color image analysis system. Images through the Wald-Wolfovitz non-parametric test of bi- were captured with a video camera, digitized, and pro- nary sequences (Siegel 1956). Correlation analysis was cessed by the NewAg386 software (Decagon Devices, used to evaluate the association among the densities of Pullman, WA, USA). seeds and adult plants between different years; we used the Dunn-S&ida´k method to obtain a corrected p for these non-independent comparisons (Sokal and Rohlf 1995). Are seed and adult densities of Camissonia dentata correlated across subsequent years? We followed adult-plant and seed-bank appearance of a common gap species on 200 permanent plots over three Results consecutive years. The species chosen was the one most Are seeds of the matrix species uncommon in the frequent in gaps (Table 1): Camissonia dentata (Cav.) soil bank of the gaps? Reiche ssp. dentata Raven (=Oenothera contorta var. di6aricata; Oenotheraceae). It is an annual herb that For the most part, yes (Table 2). The overwhelming germinates either during the southern hemisphere late majority of the seeds present in the gaps belonged to autumn or early spring (i.e. May–June or August–Sep- two annuals (Camissonia dentata and Microsteris gra- tember), and completes its reproduction by early sum- cilis [Dough. Ex Benth.] Greene) and to one spreading mer (late December–February) (Golluscio and Sala geophyte (Calceolaria polyrhiza Cavanilles). Senecio 1993). ﬁlaginoides and Bromus pictus contributed B0.1% of In January 1992, two permanent transects were es- the seeds germinating from these gap-only soil samples. tablished in the north-south direction, of 100 m each The only matrix species with a sizeable number of seeds and 5 m apart, marking at regular intervals patches of in the gaps was Poa ligularis. Overall, maximum seed-

ECOGRAPHY 25:3 (2002) 339 Table 3. Spatial distribution of viable seeds within Patagonian gaps. Samples were taken to a depth of 3 mm at 20 randomly-selected gaps, half towards the edge of the gap and half at its center. Asterisks indicate species found in the seed bank of the gaps (cf. Table 2). Different letters indicate signiﬁcant differences between positions within species (pB0.05).

Species Density (seeds dm−2) Frequency Edge Center Edge Center

Bromus pictus+B. setifolius 12.9 a0.7 b 0.75 a 0.25 b Stipa speciosa+S. humilis 8.7 a 0.5 b 0.75 a 0.15 b Mulinum spinosum 64.4 a0.5 b 0.95 a 0.10 b Senecio ﬁlaginoides 20.9 a 0 b 0.80 a 0b Adesmia campestris 25.8 a 0 b 0.60 a 0 b Poa ligularis*+P. lanuginosa 46.7 a5.3 b 1.00 a 0.45 b Hordeum comosum 8.6 a 0.1 b 0.25 a 0.05 a Camissonia dentata* 1.2 a1.9 a 0.35 a 0.40 a Calceolaria polyrhiza* 0.2 a 0.6 a 0.15 a 0.10 a Microsteris gracilis* 0.5 a0.2 a 0.10 a 0.10 a Doniophyton patagonicum 0.2 a 0 a 0.10 a 0a Cerastium ar6ense 0a0.1 a 0.05 a 0a Unknown 2.7 a 5.5 a 0.60 a 0.55 a Total 188.3 a 6.7 b 1.00 a 0.60 b bank density was recorded in summer samples (Janu- species, and Camissonia had a wing loading very similar ary), at the peak of the seed rain (or shortly after it, as to that of other species absent from the seed bank (Fig. in April 1993); very low values were found in other 1). seasons (Table 2). Several traits can be considered to be suitable estima- tors of the diaspore size: diameter, weight and area. All the correlations among them were positive and statisti- B Are seeds of the matrix species more abundant cally signiﬁcant (p 0.01; weight vs diameter: Spear- towards the edges of the gaps? man’sr=0.65, weight vs area: r=0.86, area vs diameter: r=0.75). However, none of these traits was Yes; most species were more common in soil samples correlated with the square root of the wing loading taken at the edge of the gaps, compared to samples (p\0.05). This shows that wind mobility is not neces- taken at the center of the gaps. This was true for all ten sarily related to diaspore size. matrix-forming species (Table 3). The only three species whose seeds were as frequent and abundant in the edge as compared to the center of the gaps were Camissonia dentata, Calceolaria polyrhiza, and Microsteris gracilis. These were also three of the most frequent species growing in the center of the gaps (Table 1).

Do species in the seed bank have a distinctive diaspore morphology? Yes; the most common species in the seed bank of the gaps are among the ones with the smallest diaspores in the community. The diaspores of these six species (those in Table 2 minus Bromus and Senecio) had a lower mean diameter (F=24.0, DF=1,13; pB0.001, log-transformed data), than those of the nine matrix Fig. 1. Morphological features of the diaspores of Patagonian species absent from the gaps (the ten species listed species. The square root of wing loading is inversely related to under Site description minus Poa ligularis). The two wind mobility (see main text). Values are means9SE. Aster- groups also differed – on average – on their wing isks indicate those species found in the seed bank of the gaps (cf. Table 2). Adca: Adesmia campestris; Brse: Bromus seti- loading: gap species had, on average, a larger folius; Brpi: Bromus pictus; Cade: Camissonia dentata; Capo: weight:area ratio than matrix species (F=4.7, DF= Calceolaria polyrhiza; Gila: Gilia laciniata; Hoco: Hordeum 1,13; p=0.048, square-root data). In principle, this comosum; Migr: Microsteris gracilis; Musp: Mulinum would mean that gap species may be ill-suited to be spinosum; Poan: Polemonium antarcticum; Pola: Poa lanuginosa; Poli: Poa ligularis;Seﬁ: Senecio ﬁlaginoides; Sthu: Stipa moved by wind; however, wing loading was far more humilis; Stsp: Stipa speciosa. Calculated from data in Ap- variable among gap species than among matrix-forming pendix 1. Notice log x-axis.

340 ECOGRAPHY 25:3 (2002) Are seed and adult densities of Camissonia species provides further support of the hypothesis. This dentata correlated across subsequent years? is so because we not only find the just described seed distribution among the matrix species, but also found Yes; density of the common colonizer Camissonia den- that diaspores of a different morphology were dis- tata in a given gap was positively correlated among tributed differently. A small subset of gap-dwelling years and life stages. All ten possible pairs between species is able to thrive in the mostly bare spots: seed-bank density in 1992 and 1994 (Camissonia-only Camissonia dentata, Calceolaria polyrhiza, and Micros- samples), and adult-plant density in 1992, 1993, and teris gracilis were uniformly distributed as seeds within \ 1994 yielded significant correlations (r 0.37; n=200; each gap (Table 3), and also were among the most B corrected-p 0.05). In addition, the number of gaps in successful gap colonizers (Table 1). We suggest that this which Camissonia standing plants were either consis- association was not accidental, and that the two facts tently present or absent during all three consecutive are causally connected. By contrast with what hap- years was significantly greater than expected by chance pened for nine out of ten matrix-forming species, the x2 B ( =164.9; p 0.001). colonizing ability of these three species (and also of the rare Polemonium antarcticum Gris. and Gilia laciniata Ruiz and Pavoń) could be attributed to the possession Discussion of small, very probably stagnant, diaspores (cf. Cham- bers et al. 1991). Poa ligularis diaspores, just slightly Our data strongly suggest that the recruitment of all the larger, also were abundant in the center of the gaps species forming the high-cover matrix of this Patago- (Table 1); however, they would not belong to the ‘‘gap nian community, including the dominant shrubs and specialist’’ category because these diaspores were most tussock grasses, is severely limited by the availability of abundant towards the edges of the gaps (some ten times seeds at the center of the gaps between established more; Table 3). individuals. As hypothesized by Soriano and Sala Wing loading was not as good a predictor of coloniz- (1986), this would cause a mismatch between the loca- ing ability as diaspore size was. For example, Camisso- tion of seeds and the location of safe-sites – i.e. the low nia and Calceolaria – two of the ‘‘top’’ colonizers – competition spots where soil-resource levels are less had a low wing loading; thus, all else being equal, they limiting than in closer proximity to adults (Aguiar et al. would have a tendency to be transported by wind 1994). These new results extend the findings of Aguiar comparable to that of most non-gap species (Fig. 1). and Sala (1997) and show that what they established However, another important difference between gap for the model species Bromus pictus could also be colonizers and matrix-forming species is the height applied to most of the other matrix-forming species in from which primary dispersal occurs. The seed-bearing this community. stems of all six gap dwellers are rarely taller than 10 Diaspores of matrix-forming species are nearly ab- cm, whereas for the ten matrix species this height is at sent from the seed bank of the gaps, more common least twice as much (unpubl.). It is very likely that towards the gap edges, and, because of their low wing during primary dispersal, because of this difference in loading, would have a propensity to be blown away height, matrix-forming species reach more gaps than after primary dispersal by the strong Patagonian winds. gap species do. The effect of short stature, of course, Only diaspores smaller than ca 2 mm and weighing up would be greater for gap species of higher wing loading, to 1 mg were found in the seed bank of gaps (cf. such as Polemonium, Gilia and Microsteris (Fig. 1); but Appendix 1). Several studies in wetter environments even Camissonia and Calceolaria, with a wing loading have shown that small (B1–3 mg) and compact seeds comparable to that of most matrix species, would be tend to persist in the soil longer than big and less-spher- disfavored in this respect. ical ones (Thompson et al. 1993, McDonald et al. 1996, We suggest that the syndrome including: a) the ab- Funes et al. 1999; but see Leishman and Westoby sence of dispersal enhancing traits, known as atelechory 1998). This would suggest the intriguing possibility that (Ellner and Shmida 1981), b) low stature, and c) small gap-dwelling Patagonian species are more likely to diaspore size, might to a certain point be adaptive in form a persistent seed bank, whereas species with large, windy environments. It would allow seeds to occupy highly elongated diaspores (all grasses but Poa spp.; the center of the gaps, and in the ensuing generation to Appendix 1) would not. re-occupy them. Such tendency would produce the Undoubtedly, the best way to test our main hypothe- same type of between-year correspondence that we sis – that uneven seed distributions are caused by observed in our Camissonia permanent plots. But the diaspore morphology – would be to experimentally possession of these three traits would also have a modify diaspore location and/or morphology (e.g. drawback, because it limits the population’s ability for severing or adding appendages). We did not perform mid-distance dispersal and for colonizing ‘‘new’’, neigh- such experiments or record seed movements. However, boring gaps. For example, for a dispersal height of 10 our comprehensive consideration of all the common cm, a wind velocity of 28 m s−1, and using the pub-

ECOGRAPHY 25:3 (2002) 341 Fig. 2. Density of Camissonia dentata for January 1994 in gaps along our permanent transects. These data present highly-signiﬁcant spatial aggregation (pB0.0001).

lished relationships for other types of diaspores because their diaspore morphology makes them prone (Augspurger and Franson 1987, Sipe and Linnerooth to be blown away by the strong prevalent winds. In 1995), the largest dispersal distance of Camissonia contrast, and under the same set of environmental would be between 3 and 13 m. In our case, this would constraints, a group of species with different adult size represent a distance of only between 5 and 20 gaps and diaspore morphology is able to colonize those along the wind direction. gaps. Our Camissonia permanent-plot results not only Acknowledgements – This work was funded by CONICET, show that high- (and low-) density gaps are the same Fundacioń Bunge & Born, and the Univ. of Buenos Aires (UBA). AJB held a UBA fellowship. We thank INTA for year after year, but also that these gaps display spatial permission to use its R´ıo Mayo Experimental Field. Paula patterning. Rather than being distributed at random, Amaya made the tedious measurement of diaspore traits. We high-density gaps form distinct patches separated by thank B. Boeken, E. Chaneton, M. Fenner, D. Goldberg and M. Oesterheld for comments on earlier version of this paper, low-density corridors aligned in the wind direction (Fig. and M. L. Bolkovic, I. Colonna, J. Mercau, M. Nogue´s-Loza, 2). Although it is tempting to relate this pattern to the P. Roset, and J. Vrsalovic for competent field aid. limited primary dispersal just discussed, two prelimi- nary observations indicate that such spatial arrange- ment would not be solely determined by References intra-population processes. We have found that Mi- Aguiar, M. R. and Sala, O. E. 1997. Seed distribution con- crosteris gracilis, other common gap colonizer (cf. strains the dynamics of the Patagonian steppe. – Ecology Table 1), also tend to occupy those same Camissonia 78: 93–100. patches, and at the same time to be absent from the Aguiar, M. R., Soriano, A. and Sala, O. E. 1992. Competition and facilitation in the recruitment of seedlings in a Patago- non-Camissonia ones (data not shown). In addition, the nian steppe. – Funct. Ecol. 6: 66–70. sand content of the gap soil was positively correlated Aguiar, M. R., Soriano, A. and Sala, O. E. 1994. Competition, with the maximum observed density of Camissonia facilitation, seed distribution and the origin of patches in a B Patagonian steppe. – Oikos 70: 26–34. during the 1992–94 period (r=0.22, p 0.001). Alippe, H. A. and Soriano, A. 1978. La poblacioń de dise- These last results – which warrant future experimen- m´ınulos en el suelo de un pastizal de Stipa en el oeste del tal work – point towards the existence of micro-envi- Chubut. – Ecolog´ıa3:133–138. Augspurger, C. K. and Franson, S. E. 1987. Wind dispersal of ronmental differences that could be related to the artificial fruits varying in mass, area, and morphology. – seed-trapping capacity of the gaps and/or its water Ecology 68: 27–42. balance and its influence on seedling survival (e.g. Bisigato, A. J. 1994. Dina´mica de la vegetacioń en los claros de la estepa patagońica: requerimientos de germinaciońy Coffin and Lauenroth 1994, Lauenroth et al. 1994). naturaleza del banco de semillas de Oenothera contorta var. Still, we have shown evidence of an important role of di6aricata. – Honor’s thesis, Fac. Agron., Univ. Buenos plant factors in the colonization process. Dominant Aires. Boeken, B. et al. 1998. Annual plant community responses to species of this steppe fail to colonize gaps because of a density of small-scale soil disturbances in the Negev Desert lack of properly-placed seeds, and this would occur of Israel. – Oecologia 114: 106–117.

342 ECOGRAPHY 25:3 (2002) Chambers, J. C. and MacMahon, J. A. 1994. A day in the life Leishman, M. R. and Westoby, M. 1998. Seed size and of a seed; movements and fates of seeds and their implica- shape are not related to persistence in soil in Australia tions for natural and managed systems. – Annu. Rev. in the same way as in Britain. – Funct. Ecol. 12: 480– Ecol. Syst. 25: 263–292. 485. Chambers, J. C., MacMahon, J. A. and Haefner, J. H. 1991. Matlack, G. R. 1987. Diaspore size, shape, and fall behavior Seed entrapment in Alpine ecosystems; effects of soil parti- in wind dispersed plant species. – Am. J. Bot. 74: 1150– cle size and diaspore morphology. – Ecology 72: 1668– 1160. 1677. McDonald, A. W., Bakker, J. P. and Vegelin, K. 1996. Coffin, D. P. and Lauenroth, W. K. 1994. Successional dy- Seed bank classification and its importance for the restora- namics of a semiarid grassland; effects of soil texture and tion of species-rich flood-meadows. – J. Veg. Sci. 7: 157– disturbance size. – Vegetatio 110: 67–82. 164. Crawley, M. J. 1990. The population dynamics of plants. – Mott, J. J. and McComb, A. J. 1977. Patterns in annual Philos. Trans. R. Soc. Lond. B 330: 125–140. vegetation and soil microrelief in an arid region of western Ellner, S. and Shmida, A. 1981. Why are adaptations for Australia. – J. Ecol. 62: 115–126. long-range seed dispersal rare in desert plants? – Oecologia Paruelo, J. M. and Sala, O. E. 1995. Water losses in the 51: 133–144. Patagonian steppe: a modelling approach. – Ecology 76: Eriksson, O. and Ehrleń, J. 1992. Seed and microsite limitation 510–520. of recruitment in plant populations. – Oecologia 91: 360– Paruelo, J. M. et al. 1991. The use of satellite imagery in 364. quantitative phytogeography: a case study of Patagonia. – Fenner, M. 1985. Seed ecology. – Chapman and Hall. In: Nimis, P.-L. and Crovello, T. J. (eds), Advances in Fernańdez, R. J., Sala, O. E. and Golluscio, R. A. 1991. phytogeography. Kluwer, pp. 183–204. Woody and herbaceous aboveground production of a Paruelo, J. M. et al. 1998. The climate of Patagonia: general Patagonian steppe. – J. Range Manage. 44: 434–437. patterns and controls on biotic processes. – Ecol. Aust. 8: Fernańdez, R. J., Nun˜ez, A. H. and Soriano, A. 1992. Con- 85–102. trasting demography of two Patagonian shrubs under dif- Reichman, O. J. 1984. Spatial and temporal variation of seed ferent conditions of sheep grazing and resource supply. – distributions in Sonoran Desert soils. – J. Biogeogr. 11: Oecologia 91: 39–46. 1–11. Funes, G. et al. 1999. Seed size and shape are good predictors Sala, O. E. et al. 1989. Resource partitioning between shrubs of seed persistence in soil in temperate mountain grasslands of Argentina. – Seed Sci. Res. 9: 341–345. and grasses in the Patagonian steppe. – Oecologia 81: Garner, W. and Steinberger, Y. 1989. A proposed mechanism 501–505. for the formation of ‘‘Fertile Islands’’ in the desert ecosys- Schlesinger, W. H. et al. 1996. On the spatial pattern of soil tem. – J. Arid Environ. 16: 257–262. nutrients in desert ecosystems. – Ecology 77: 364–374. Golluscio, R. A. and Sala, O. E. 1993. Plant functional types Siegel, S. 1956. Non parametric statistics for the behavioral and ecological strategies in the Patagonian forbs. – J. Veg. sciences. – McGraw-Hill. Sci. 4: 839–846. Sipe, T. W. and Linnerooth, A. R. 1995. Intraspecific variation Golluscio, R. A., Leoń, R. J. C. and Perelman, S. B. 1982. in samara morphology and flight behavior in Acer saccha- Caracterizacioń fitosociolo´gica de la estepa del Oeste de rinum (Aceraceae). – Am. J. Bot. 82: 1412–1419. Chubut; su relacioń con el gradiente ambiental. – Bol. Soc. Sokal, R. R. and Rohlf, F. J. 1995. Biometry, 3rd ed. – Arg. Bot. 21: 299–324. Freeman. Greene, D. F. and Johnson, E. A. 1986. Dispersal capacity Soriano, A. 1960. Germination of twenty dominant plants in and seed production in anemochorous plants. – Oecologia Patagonia in relation to regeneration of vegetation. – In: 68: 629–631. Proc. Eight Int. Grassl. Congr., pp. 154–158. Greene, D. F. and Johnson, E. A. 1993. Seed mass and Soriano, A. 1983. Deserts and semideserts of Patagonia. – In: dispersal capacity in wind dispersed diaspores. – Oikos 67: West, N. E. (ed.), Temperate deserts and semideserts. 69–74. Elsevier, pp. 423–460. Grubb, P. J. 1986. Problems posed by sparse and patchily Soriano, A. and Sala, O. E. 1986. Emergence and survival of distributed species in species-rich plant communities. – In: Bromus setifolius seedlings in a Patagonian steppe. – Isr. J. Diamond, J. and Case, T. J. (eds), Community ecology. Bot. 35: 91–100. Harper and Row, pp. 207–225. Soriano, A., Sala, O. E. and Perelman, S. B. 1994. Patch Klausmeier, C. A. 1999. Regular and irregular patterns in structure and dynamics in a Patagonian arid steppe. – semiarid vegetation. – Science 284: 1826–1828. Vegetatio 111: 127–135. Lauenroth, W. K. et al. 1994. The importance of soil water in Thompson, K., Band, S. R. and Hodgson, J. G. 1993. Seed the recruitment of Bouteloua gracilis in the shortgrass size and shape predict persistence in soil. – Funct. Ecol. 7: steppe. – Ecol. Appl. 4: 741–749. 236–241.

ECOGRAPHY 25:3 (2002) 343 Appendix 1. Diaspore weight and size for 16 common Patagonian species. Values are mean and coefﬁcient of variation (CV %); sample size=10. Asterisks indicate species found in the seed bank of the gaps (Table 2).

Species Life form Weight (mg)Area (mm2) Length (mm) Width (mm) Length/width (ordered by weight) mean CV mean CV mean CV mean CV

Calceolaria polyrhiza* Herb 0.07 8.6 0.389.9 0.78 7.5 0.45 7.5 1.73 Camissonia dentata* Annual 0.08 9.9 0.46 8.1 0.963.7 0.46 6.3 2.09 Gilia patagonica* Annual 0.40 8.9 0.848.9 1.22 5.0 0.72 3.8 1.69 Poa ligularis* Grass 0.52 7.7 2.86 9.7 3.78 6.2 0.97 7.3 3.90 Senecio ﬁlaginoides Shrub 0.56 27.7 28.34 25.5 3.26 8.9 0.58 22.4 5.62 Poa lanuginosa Grass 0.63 16.3 4.4119.0 4.81 8.8 1.50 25.9 3.21 Microsteris gracilis* Annual 0.83 9.41.77 8.0 1.76 6.8 1.10 3.9 1.60 Polemonium antarcticum* Annual 1.06 4.6 1.276.5 1.70 3.5 0.83 5.4 2.05 Stipa humilis Grass 3.47 26.8 43.67 59.1 61.36 6.6 0.91 16.2 67.43 Hordeum comosum Grass 4.25 26.9 25.90 20.8 20.61 13.1 7.23 32.1 2.85 Mulinum spinosum Shrub 5.44 13.2 19.7518.5 5.51 9.7 4.64 10.0 1.19 Bromus setifolius Grass 5.85 14.3 40.32 13.7 16.60 7.1 3.40 8.1 4.88 Bromus pictus Grass 7.24 5.5 36.783.7 14.36 2.0 3.61 2.5 3.98 Adesmia campestris Shrub 7.49 17.4 50.65 27.6 14.46 12.710.01 11.3 1.44 Stipa speciosa Grass 7.91 15.1 46.46 56.3 73.44 34.21.21 11.2 60.69

344 ECOGRAPHY 25:3 (2002)