Granivory in Southern South American Deserts: Conceptual Issues and Current Evidence

Ecologists have abandoned attempts to develop a simple, unifying theory to account for community patterns (Lawton 1999). Instead, to explain such patterns, they rely on local ecological processes as well as on more contingent historical processes acting at large temporal and spatial scales (e.g., unique historical events controlling speciation and biogeographic interchanges). This mixed approach should lead to a more realistic idea of the real world, but it deserves careful epistemological consideration. Although historical hypotheses may account for a substantial part of community patterns, they impose limits to theory development because they deal with contingent events that may often be theoretically intractable. An ecological system may be unique, but this does not necessarily imply that it is lawless in the ontological sense. Consequently, the methodological problem remains of distinguishing its general properties from its idiosyncratic properties (Mahner and Bunge 1997). Historical explanations also impose limits to hypothesis testing because they often do not make predictions and can be elucidated solely by historical reconstruction (Schluter and Ricklefs 1993). Because theory development is a prerequisite for ecological understanding, researchers should continue to develop general theories, even as they recognize the importance of history (Pickett et al. 1994, Mahner and Bunge 1997). Theory development can be fostered by testing the predictions of the “community convergence” hypothesis through comparative studies (Schluter and Ricklefs 1993). This hypothesis specifies that independently assembled communities in similar but geographically distant habitats converge in composition and functioning if they face similar environmental pressures (i.e., if they are molded by the same local ecological processes). Ecologists have long searched for matching patterns in the biota of arid lands because similar harsh pressures are thought to prevail in deserts (e.g., Orians and Solbrig 1977, Mares 1993a). An example of such an analysis is the set of experimental tests for convergence in seed harvesting by granivorous mammals, ants, and birds in deserts worldwide (Mares and Rosenzweig 1978, Abramsky 1983, Morton 1985, Kerley 1991, Vásquez et al. 1995, Lopez de Casenave et al. 1998). The initial conclusions of such experiments indicated that seed removal by mammals was higher in deserts of the northern continents than of the southern continents, that ants are the main seed harvesters in southern deserts, and that seed removal by birds is low in deserts around the world (Mares and Rosenzweig 1978, Abramsky 1983, Morton 1985, Kerley 1991, Kerley and Whitford 1994). Granivory may have a significant community role in the monte and other south american deserts, despite the probable lack of highly specialized seedeaters These studies seem to suggest a lack of convergence in seed harvesting. However, comparative studies depend strongly on the robustness of established patterns, and the above generalization was not tested for robustness but inferred by using one research approach alone (i.e., bait removal experiments). Furthermore, this generalization was supported, in some cases, by only one experiment carried out in one location within a particular desert. Therefore, some of the purported patterns of granivory at the continental scale might be artifacts of the limited basis of empirical information (Mares 1993a). For example, Mares and Rosenzweig (1978), Abramsky (1983), and Morton (1985) all asserted that total granivory in the extensive Monte Desert of Argentina is depressed, and granivorous assemblages depauperate, in comparison with other warm temperate deserts. However, this conclusion was derived from a single experiment carried out near Andalgalá, in the far northern reaches of the Monte Desert (Figure 1; Mares and Rosenzweig 1978). Despite the narrow domain of this study, many scientists have tacitly extended its conclusions to arid South America as a whole (Abramsky 1983, Morton 1985, 1993, Brown and Ojeda 1987, Kerley 1991, Wurm 1998; but see Vásquez et al. 1995, Lopez de Casenave et al. 1998). The Monte Desert in perspective. (top) The world's desert regions, including some semi-arid areas that are not generally considered true deserts, are indicated by shading. (inset) The Monte Desert in southern South America is a relatively narrow north–south belt that is oriented along the eastern slopes of the Andes Mountains. It broadens to the south and extends to the Atlantic coast in northern Patagonia. Seed removal rates by granivorous animals have been assessed in three localities, all in Argentina: A, Andalgalá, Catamarca Province; Ñ, Ñacuñán, Mendoza Province; P, Puerto Madryn, Chubut Province. In this article, we summarize current evidence about the resource base of granivorous animals, the pressures these animals impose on seed reserves, and the main ecological features of the assemblages of seed-eating ants, mammals, and birds in the Monte Desert and other southern South American deserts. We examine the general validity of the commonly accepted patterns of granivory by using multiple research approaches to verify pattern robustness. If the patterns are robust, they should be the basis for more reliable tests of the community convergence hypothesis of granivorous assemblages in deserts from different continents. Such tests could provide a more informed assessment of the relative roles of local ecological processes and unique historical circumstances in organizing desert communities. Mares and Rosenzweig (1978) observed that “total granivory is much depressed in the Monte Desert, where granivorous mammals are rare and ill-adapted, ants are depauperate and not usually granivorous, and birds are unimportant seed consumers,” and they suggested that low granivory would be the consequence of a seed decline in South American deserts caused by the relatively recent extinction of the marsupial family Argyrolagidae in South America. Mares and Rosenzweig (1978) assumed that argyrolagids were ecological equivalents of the rodents of the North American family Heteromyidae (Sánchez-Villagra and Kay 1997), and they outlined two alternative evolutionary scenarios whereby marsupial extinction might have led to a decline in granivorous assemblages (particularly of ants) via a decrease in their resource base (i.e., seeds). In the first alternative, argyrolagids were indirect evolutionary mutualists with ants, consuming seeds not preferred by ants. Thus, when the argyrolagids became extinct, ant-preferred plants, whose seeds were still consumed, would have been outcompeted by predator-free, ant-avoided plants. The decreased availability of seeds from ant-preferred plants would have led, in turn, to a decline in the granivorous ant assemblages themselves. In the second scenario, argyrolagids were mutualists with food plants shared with ants, perhaps storing some of the seeds of these plants in surface caches and thereby enhancing germination, as has been shown among North American heteromyid rodents (Mares and Rosenzweig 1978). After the extinction of argyrolagids, those plants would have suffered a decrease in their populations, and the ant assemblages would also have declined in abundance and richness. Although these historical explanations about the role of extinct argyrolagids cannot be tested directly, they do lead to testable predictions about the status of present-day seed reserves in the Monte Desert (Mares and Rosenzweig 1978, Brown and Ojeda 1987). For example, the seeds consumed by present-day ants, as well as those presumably involved with marsupials in the past, are predicted to be less abundant in seed banks of South American than North American deserts (Marone and Horno 1997). We tested this assertion by comparing seed reserves in woodland and shrubland of the central Monte Desert with those in several other South and North American deserts (Marone and Horno 1997). Densities of seeds in soil banks were similar in arid North and South America at the habitat scale (Table 1). Regardless of continent, most reports fall within the range of 8,000–30,000 seeds/m2 proposed by Kemp (1989), with the exception of one extraordinarily high value in the eastern Mojave Desert of California (Price and Joyner 1997). a Average total seed number and seed mass in soil seed banks in different habitats from several South and North American warm deserts.b a Average total seed number and seed mass in soil seed banks in different habitats from several South and North American warm deserts.b Moreover, with respect to production of seeds in South America, the few available data suggest seed production rates similar to those in North America. By using seed traps to catch the seed rain during primary dispersal, we estimated grass seed production in years of moderate rainfall in the central Monte Desert to be 12.5 kg/ha (Marone et al. 1998a). Using similar methods, Pulliam and Parker (1979) and Pulliam and Dunning (1987) reported average grass seed production of 13.0–15.8 kg/ha in moderately rainy years in grasslands of the North American Chihuahuan Desert of southeastern Arizona. Finally, seed numbers at the microhabitat scale as well as the proportion of small seeds (i.e., those presumably preferred by ants) also appear to be similar in deserts of North and South America (Marone and Horno 1997). Overall, these findings fail to support the hypothesis of a seed decline in the South American sites. Therefore, differences in the resource base should not be invoked to account for differences in granivorous assemblages (Mares and Rosenzweig 1978) or in the foraging behavior of particular species (Medel and Vásquez 1994) between South and North American deserts. Morton (1985, 1993) also ruled out differences in the resource base as an explanation for differences in granivorous assemblages between Australia and North America. The results of Mares and Rosenzweig (1978) on present-day consumption of seeds in the northern Monte Desert by different granivorous taxa in bait removal experiments are still often extrapolated to the entire Monte Desert (e.g., Wurm 1998). To assess the applicability of those results to the central Monte Desert (Figure 2), we the same of in seed consumption is estimated by comparing the of seeds by different taxa from in the desert (Figure can be in comparing results from bait removal experiments because such experiments often lack For different have a of seeds to and have seed among in even the of seed consumption by mammals, and ants may with the of seed et al. or the microhabitat de Casenave et al. 1998). Consequently, researchers to when comparing seed removal rates et al. et al. Vásquez et al. The woodland in the of Ñacuñán, in the central Monte Desert, We carried out in and in of the between and is and with and most rainfall in of the and are abundant in woodland and seed production and are to and (Marone et al. 1998a). in bait removal experiments de Casenave et al. 1998). of seeds is in at the of the of three in a a and an are to the of long and set above the to by ants. for birds are available only during the and seeds for mammals are available only at are with the of the with the soil surface and with to by a The are with of seeds to of the experiment is when we the most of seed consumption (i.e., the average seed removal in in seed consumption by birds and ants in the was than in the of and we that the total of seed consumption in the central Monte de Casenave et al. was more than than in the northern Monte (Mares and Rosenzweig 1978). if data from the central Monte Desert are of the of desert on seed the most conclusion to be that total granivory in the Monte is not depressed de Casenave et al. total seed removal as well as seed removal by mammals and ants is high in North American deserts Morton To Vásquez et al. of the relative importance of mammals, and ants among bait removal experiments We summarize their findings in results from the central Monte Desert as well as an data set from the southern Monte Desert near Puerto Madryn, Argentina (Figure 1; Puerto Madryn, Although these only the relative of taxa in (i.e., they about granivory they that the relative of mammals is in northern continents and that the relative of ants is in southern continents. The also a granivorous birds are seed in several South American a of the relative importance of the three main taxa small as suggested by bait removal experiments carried out in different arid of the a of the relative importance of the three main taxa small as suggested by bait removal experiments carried out in different arid of the The importance of seed-eating birds to some of seed To assess the of the of on seed in deserts around the world, it is to seed availability and all of seed may among deserts, even when total seed remains that have far less than and ant consumption seed and and seed caused by and These and other of seed should be historical or ecological are more to explain the of the system at and scales and Whitford 1994, and Horno 1997). have often assumed that assemblages of granivorous ants are less in arid South America than in similar habitats of North America (e.g., Mares et al. 1977, Mares and Rosenzweig 1978, Brown and Ojeda 1987, and For example, Mares et al. ant communities of a Desert near and the northern Monte Desert near a higher species of ants in but granivorous at The status of ants, was not on foraging but on on the status of ants was to foraging the conclusions ant assemblages of North America, and South America by using seed removal the of Morton and ants in in the central and northern Monte Desert, including Andalgalá, and that these assemblages in more and abundant than their North American (Medel and Vásquez 1994, Using the same we assessed the composition of ant assemblages in the central Monte Desert Lopez de In 1995, we and species seeds from two sites. This species the of and reported for the northern Monte Desert and central (Medel and Vásquez as well as the of and reported for North American and deserts, and In foraging of ants is not in with that granivorous ants are depauperate in the Monte The bait removal in all these studies is to the of If ant species are considered to be seed harvesters because they are at seed the number of ant species may be By we only ants that seeds from seed Consequently, we that assemblages of ants are not depauperate, at in the central Monte The South American ant several species with more than those of the ants of the make the of ant assemblages in North American deserts. Although Mares and Rosenzweig (1978), and and Vásquez all that of the ant species in South American deserts results from a seed availability in these deserts than in North American deserts, the in do not support this species in South America as The three species and that in the central Monte Desert and 1993) appear to be highly specialized Lopez de However, that the South American species of this seem to and than those in North America and the hypothesis of a of South American species on seeds available as well as in is a still of ants and food among some other ant species might explain the Monte granivorous ants appear to seeds from experimental than do North American ants. it does not that the ants of the Monte Desert depauperate and not usually (Mares and Rosenzweig 1978). To differences in seed removal from differences in ant it be to examine more the foraging behavior of the Monte ants. For example, that some species of the Monte Desert the of their seeds from the same the of of these highly specialized seed-eating species was low at experimental seed on the soil Lopez de differences in ant foraging behavior among continents might the of results from this of the conclusions about the importance of granivorous ants as seed of seed removal from bait and of food of granivorous animals are but not to make about the community role of approach to the community role of those animals is to assess the to the of the resource (i.e., is by the of the (i.e., ants, or the importance of granivory can be through studies of the of seed consumption on the of seeds (Price and Joyner et al. The of of North heteromyid rodents and the low and abundance of other in South American deserts have been for the seed consumption in the northern Monte Desert (Mares et al. 1977, Mares and Rosenzweig 1978, Brown and Ojeda 1987). suggest that this of seed consumption does not to the entire Monte rodents an of more seeds in the central Monte Desert than in the northern Monte Desert de Casenave et al. 1998). seed removal is also higher in the southern Monte than in Thus, even if South America of the that granivory is in the Monte Desert deserves for at two the of their at and In a of several heteromyid on average their to more than of during a single in the et al. 1997). a these rodents may far more seeds than they to their Although some South American rodents have been shown to some at these species lack and It on South American rodents would be to seeds from than do indirect evidence to this of seed consumption by desert rodents South American lack were higher when on of than when on bait removal for with about granivory in South America is that information is still for South American desert and desert often about their food Although this the that the proportion of seeds in in and is also fostered by the lack of This lack of information is a problem because some of seed consumption (e.g., those on are to in such as are not in a a (e.g., a granivorous is an that some or all food in different than the at they are in the where the within the limits by the and of the Because this is to in more could be that they are only as a basis of comparative For example, Kerley and Whitford suggested that mammals can be as or when their are by the (i.e., more than of the of the those of more mixed can be Using this that two species in the southern of the Desert of and were granivorous, in the when and of their of with of its of and that in several of the Andes of southern more than of the of some species of seeds (e.g., that some heteromyid rodents that are usually considered specialized seedeaters also significant of and during some (Mares it that some of the and species should also be considered In the Monte Desert, Mares et al. and Mares reported granivorous small In some of the central Monte Desert, may be the most abundant species and data from other semi-arid habitats of Argentina that the of of seeds and 1994). In a recent in the central Monte Desert, that during the to a relatively high proportion of seeds and that the of other small mammals few seeds than in the of and However, these low may the proportion of seeds in the of and the other small indicated that the that food other than seeds in by only the remains of grass rodents often grass and the consumption de Finally, rodents seeds as well as grass For of the of in Argentina of and of it of seeds and 1994). North American species of the heteromyid and are usually as specialized (e.g., et al. they also and (Mares Kerley and Whitford 1994). Moreover, Kerley et al. have proposed that the status of in the Chihuahuan Desert might be to (i.e., than to By those South American rodents (e.g., several and that appear to similar in at some at are in the This has for the conclusions of comparative may an of research that deserves the of the of when empirical information on a particular is or may as even such are at This should when these (e.g., that the of rodents to seeds is similar in North and South America, or that the of seeds in the of South American rodents is are tested about the of the hypothesis (e.g., that seed-eating mammals in the Monte Desert and other South American deserts have a significant on seed This research is to in a hypothesis is considered by the of one of its predictions (e.g., that South American small mammals similar of seeds in bait experiments as their North American The low rates of seed removal by birds observed in bait removal experiments to suggest that birds are unimportant seedeaters in deserts or that they fail to the For example, birds few of the seeds in the South the Chihuahuan Desert et al. and the Desert the and abundant of deserts, Morton (1985, 1993) that these birds should have a community role in deserts, even they not of seed have suggested that the of bait removal experiments may be for seed consumption by desert birds (Mares and Rosenzweig 1978, Morton In most bait experiments carried out in South American deserts, and consumed birds often consumed more seeds than the other taxa (Table We of idiosyncratic of South American birds that may explain this Thus, to for of the Monte to seed with we an approach that not rely on to assess birds are seedeaters in the central Monte Desert (Marone et al. experiments have shown that are the main seed in the central Monte Desert (Figure see Lopez de Casenave et al. 1998). To assess the of seed-eating birds on the soil seed reserves, we the soil seed in the of with that the also account the total number of seeds that the soil from to (i.e., we seed et al. In consumption by the seed of grass seed banks was or The the of grass or as well as the of consumption in and of grass seeds was or and large grass seeds suffered higher than small of seeds in the seed were between or and or seed production during the been relatively low or These results suggest that seed in the central Monte Desert grass seeds rates of seed removal by in the central Monte ants, mammals, rates of removal for the available only to the are a with the same are not different to from Lopez de Casenave et al. and with from of evidence that grass seed during in the central Monte Desert is to granivorous ants are during the and (Figure Lopez de Casenave et al. 1998). Furthermore, in and and total grass seed germination, usually does not of soil seed reserves (Marone et al. only of reserves during the of was with a Finally, the composition of the seed at different and et al. that grass the and large would by Therefore, it was not that the pattern of seed with the pattern of seed consumption by granivorous birds in the and of through of the seed mass in from grass and only from Furthermore, and large grass seeds suffered the from the soil seed and were also the main of foraging by Moreover, a significant grass species between the mass decrease of seed species from the soil and the mass of those seed species in in (Figure et al. This that consumption could explain a deal of the grass seed observed in of the central Monte between the mass of seeds from in the and of granivorous species in the central Monte Desert and the of seeds of species during the and of The species are and The was significant species are as The for is not shown on the but with that for These of evidence that (particularly have a on the and of seed reserves in the Monte Desert, where granivory in and granivory in been considered Moreover, seeds from a of This the (Marone that the and of rainfall in the central Monte Desert may the abundance and of granivorous birds via the of seed production by to in between seeds and in this South American desert may be more than The relative importance of seed consumption or granivorous assemblages in deserts around the world cannot be assessed through one research approach alone (e.g., bait removal of the species composition of granivorous or of Instead, research that multiple approaches and of hypotheses and may lead to the most conclusions to and (Pickett et al. 1994). Furthermore, such should studies that account the spatial and temporal of communities. as a consequence of such a lack of rates of seed consumption and the abundance and of granivorous assemblages in some South American deserts may have been in the The evidence we have in this from desert of southern South America and was by using different research This evidence that birds and ants are seedeaters during the and and that the role of small mammals as in the central Monte Desert of Argentina deserves more These results should be comparative studies to between idiosyncratic and general processes desert communities Although ecological are by ecological continue to depend on theory development (Pickett et al. 1994, Mahner and Bunge 1997). To develop ecological researchers with and realistic experiments carried out in the of research that are on testing to account for ecological 1998). We and for their and during and Horno and and at the initial more of we are to of was to of the were through from and different and were by the and the and the of de de de and the and de and the for and Bunge for the at where the of this was This is of the Desert