Abstract. Phrynosoma modestum  and P. cornutum  coexist in a creosotebushdesert grassland community in southeastern Arizona.  Their ant diet changes with increasing lizard size.  The two species have similar diets within size categories, except that the largest P. cornutum  is different within and between species. It is suggested that feeding occurs as  haphazard trips through an environment of different ant nests with opportunistic feeding on encounters with ants of different behaviors and spatial distribution.  It is possible that the largest P. cornutum  learn to selectively feed on the largest species of harvester ant.
 

         INTRODUCTION

         In 1958 a permanent cattle grazing exclosure was established to study the  long-term dynamics of a “new ecosystem” dominated by desert shrubs that had replaced black grama grassland previously degraded by over grazing, drought and soil erosion. The vegetation was dominated by creosote bush, Larrea tridentata, with lesser coverage by subshrubs: Parthenium incanum, Zinnia pumila  and Gutierrezia sarothrae,  and fluff-grass,Tridens pulchellus.  As part of an ecosystem study, I recorded the presence and location of lizards. Five species predominated: the whip-tail lizards, Cnemidophorus tigris  and C. uniparens,  the side-blotch lizard, Uta stansburiana,  and two horned lizards, Phrynosoma cornutum  and P. modestum. Variabilities of their densities have been followed over 41 years, as will be presented elsewhere.  The sympatry of the two ant-specialist Phrynosoma invited special attention as Shaffer and Whitford (1981) suggested that their coexistence depends upon different uses of ant resources.  From 1977 through 1983 I collected individuals of both species, held them in captivity until they passed fecal pellets, and then returned them to the point of collection. Small numbers were obtained in any one year because of the low densities of the lizards and their low visibility as sit-and-wait predators,  Whitford and Bryant (1979), Shaffer (1979) and Shaffer and Whitford (1981) give foundation information on the daily behavior of P. cornutum and P. modestum, and analyses of fecal pellets.  These studies were in a desert shrub environment similar to my study site except for the composition of the vegetation.  Munger (1984 a, b, c, 1986) gives information for both species on foraging behavior, mortality and home range.  By watching with binoculars he estimated diets as lizards preyed on ants. Munger’s information is particularly relevant since his study site is only 9 km southeast of my site, but again it is in a different vegetation.

        Since there are many variables that can influence the feeding of Phrynosoma,  I think the present study is a useful extension to the previous studies. It has several differences that may enhance understanding: (1) Diets are analyzed in terms of the sizes of lizards, which has not been done thoroughly before. (2) Diet information extends over seven years, which may give a better measure of its consistency or variability, if there has not been any important change in the ant community, which is not certain. The interesting questions are: (1) Is there an important difference in the use of ant resources by the two species?  (2) Does body size affect resource use in an important way? (3) Are resources used by chance and/or in some selective way?
 

        METHODS AND MATERIALS

        Site. The study site is a 9.3 ha grazing exclosure on a gently sloping bajada on the western side of the San Simon Valley, just off the eastern foothills of the Chiricahua Mountains, 8 km N of Portal, Cochise Co., Arizona.  Vegetation, soil and climate of the site are described in Chew and Chew (1965).  The present work was done on 5.6 ha dominated by creosote bush;  this area is separated by a shallow wash from an adjacent vegetation dominated by tarbush, Flourensia cerenua, and prickly pear, Opuntia phaeacantha. The vegetation in 1958 was typical of desert scrub of the Chihuahuan Desert, but it had been semiarid grassland in the 1800s.

        Ant community,  Extensive information was gathered on the ant community in 1972 and 1976-77 by observations centered on bait boards (Chew 1977, Chew unpublished, Chew and De Vita 1980). Table 1 lists the ants that were eaten by Phrynosoma,  together with certain of their general characteristics.  Dry weights measured in 1972 are used as a proxy for the digestible substance of individual ants, although this can be confounded by different ratios of chitin to digestible substance among ant species, and more so between ants and other insects. The total density of nests for all ant species, ca. 1400/ha, is similar to that found by Gaspar and Werner (1976) for three of four sites in the Sonoran Desert (352, 1168, 1200 and 2162 nests per ha).

        Horned lizard parameters. Table 2 shows the seasonal distribution of sightings of horned lizards and of the samples of their feedings.  The monthly distributions of Phrynosoma  are like those of ant activity at bait boards, except that most ants peaked in June-July and had sharper peaks of abundance than the lizards (Chew, unpublished).  P. cornutum (hereafter Pc ) may have a more limited season of activity than P. modestum   (Pm ), emerging from hibernation later and reentering earlier.  This could be related to reproductive biology.  Howard (1974) suggests that some Pm breed in their first year and all may breed after one winter hibernation, whereas Pc probably breed only after a second hibernation.

        Fifty seven Pm  and 22 Pc  were captured and measured at various times  (Table 3).  All Pm  were within the size range of Pc, whereas 41% of Pc were larger than any Pm. The information for Pc  is limited, but it is clear that the two species have individuals of the same sizes feeding from the same ant resources, which could be important if prey use and body size are related.

         For use in analysis of ant diets, the total size range of lizards was divided into four categories: 22-36 mm snout-vent, 37-52 mm, 53-65 mm, and (only Pc) 66-103 mm.  These categories were selected on the basis of gaps in the distribution of sizes of the animals obtained, their approximate equal spans, and on the approximate equal numbers of animals in each category.  The ant species were divided into three size groups: small 0.06-0.40 mg dry wt, medium 0.66-1.05, and large 1.70-2.86.

        Howard (1974) found that Pm males were mature at snout-vent lengths >41 mm.  Pianka and Parker (1975) give a snout-vent of 50 mm for mature Pm and 69 mm for Pc, and maximum sizes of 60 mm (males) and 67 mm (females) for Pm  and 94 mm and 114 respectively for Pc .

         Analysis of fecal pellets. Lizards captured in the field were held in the lab until they passed fecal pellets. Each Pm   produced one to four pellets (median 2-3. total 24 pellets). Pc   also produced one to four pellets (median 3, total 15). Each pellet is here assumed to represent a different feeding period in the field, although this may not always be true. For Pc   there was a significant relationship between snout-vent size of a lizard and its maximum pellet length:

The regression was not significant for Pm  because scat lengths for small animals were all outliers from the fitted regression.  Six pellets collected fresh in the field were of such size (22-32 mm) that they exceeded the maximum size that could have been produced by Pm (est 19 mm), and these were analyzed as feedings of Pc.

        Pellets were broken up and the ant heads they contained were identified without difficulty by use of reference collections of whole mounted ants, except that because of the variability of size of their heads, minors of Pheidole rugulosa and Ph. xerophila could not be discriminated. When majors were present, the minors were assigned to that species.  In tables these two species are combined as "Ph. rug-xer".  The tallies were analysed in terms of total items and the estimated weights of items for each species.  Remains of insects other than ants were identified by comparison to specimens in collections of ground-dwelling arthropods.  Dry weight was estimated from that of whole specimens in the collections, with proportional adjustment for size differences.

        Similarities of feedings were calculated for comparisons of composition within the several feedings of individual lizards, and betweed lizards in each size class. Similarity was calculated as

 where p is the proportion of species i as it is at a minimum in the feedings of lizard species j or k.

        Feeding niche overlap was calculated for comparisons of size classes within and between the two Phrynosoma, using the index of Pianka (1973):

        RESULTS

        Diet summaries. Table 4 summarizes the importance values:

for each ant species, IV by numbers and IV by weight, in the feedings of each  size class of P. modestum and P. cornutum.  Importance value is used here because it makes an adjustment for the frequency of species in feedings.

        Table 5 shows the average total number and total dry weight of all ants in feedings of each size-class of lizards.  There is no significant difference between lizard species within size categories.  However, the weight of ants is significantly different or near significant between size categories within Pc, and between Pm of 22-36 mm and larger size classes.

        Table 6 shows the frequency with which each ant species is present in the feedings of each size class of lizards.  In some cases there may be important changes in the importance value of an ant species from one size to another, without any change in the frequency of presence of the species in the diet, i.e. a change of numbers but not frequency.

        Ant profiles. In many instances there is a large range in the number of individuals of an ant species in the feedings of a size class of lizards, the extreme being 1-759 for Ph. rug-xer  in 66-103 mm Pc. Table 7 shows that all but one species of ant, on one or more occasions,  provided 50% or more of the total weight in a feeding by one or both Phrynosoma.  In Table 4, the large incidence of Ph. rug-xer in some feedings of 53-65 and 66-103 mm Pc were adjusted downward (to equal the average for other feedings on Ph. rug-xer  in that size class) so that importance values for other ant species would not be distorted.  This adjustment was not made for the averages in Table 5.

        Diet comparisons. Table 8 shows the similarity of feedings within samples from the same individual lizard versus the similarity between randomly chosen feedings of different individuals in the same size class.

        Table 9 shows the diet niche overlaps between size classes within species and between the two lizard species.

        Diet items other than ants. Fifty percent of fecal pellets of Pm and 53% of those of Pc  contained insects other than ants.  Table 10 summarizes the incidence of these insects.

        The variation that can be present in a small number of samples spread over seven years imposes caution on speculation.  However, in this study there is no evidence for a major difference in ant resource use, like that found by Shaffer (1979) and Shaffer and Whitford (1981). They concluded that for their site P. modestum and P. cornutum “can coexist because of differences in prey selection”, modestum having Myrmecocystus depilis and M. mimicus as its most dependable prey and cornutum depending on a combination of Pogonomyrmex  spp., the result of differences in foraging behavior that lead modestum and cornutum  to encounter ant species differently.  At a certain point, Pm  went into shade and there fed on Myrmecocystus as they also moved into shade and went up into mesquite to feed on insects and honeydew. Myrmecocystus averaged 24.7 individuals in fecal pellets, out of an average of 70.9 ants (34.8%). P. cornutum did not have this behavior.  Munger 1984b) also found that Pm  took Myrmecocystus  in greater proportion (40% of identified prey captures) than Pc (11%). This was for visual observation of 14 Pm  and 11 Pc, each followed for one day.

        M. depilis contributed much less to the feedings of Pm at my site: overall 8.5% of 3547 individuals, although the density of their nests, 22.2/ha, was the same as Shaffer (1979), 21.7/ha. Possibly this is because the principal shade plants, cresoste bushes, provide less food for M. depilis than mesquites, which are a negligible part of plant cover.  This suggests that the vegetation of an ecosystem affects the ant resources and hence their use by Phrynosoma.  At the present site ant samples taken in the area dominated by Larrea and that  dominated by Flourensia and Opuntia had a similarity of abundances of only SI 0.36 (Chew, unpublished). Most of the similarity (51.2%) was due to I. pruinosum; other species were quite dissimilar. Gaspar and Werner (1976) compared Sonoran Desert sites and concluded, “In a very general way, and from a quantitative viewpoint, the ant populations differ significantly from one site to another.”

        In this study the two Phrynosoma  species are united in their high usage of Po. imberbiculus .  This ant was present in 85-89% of the fecal pellets of Pm  and 92-100% of those of Pc; Po. imberbiculus  ranked 1st in importance by weight in all size classes of both Phrynosoma, except the largest 66-108 mm Pc . In their feedings the dominance of Po. imberbiculus is abruptly replaced by that of the larger Po. desertorum (IV numbers = 0 500, wt = 0.713, Table 4).  This may be an important adaptation for Pc since it achieves sexual maturity at these sizes, and large ants would more effectively supply the additional nutrients needed for
reproduction than a greater number of smaller ants.  By contrast, Pm, which are sexually mature at 53-65 mm, are feeding on the same ants as Pc of that size.

        There is no major difference in weight intake of different ant species by the two horned lizards within their three common size categories (Table 4).  Other reports on Pm and Pc sympatry are that: (1) “paired comparisons for ant utilization by size show  that dietary overlap is 0.25” (Barbault and Maury 1981), but there are no illustrative details. (2) Munger (1984a) found substandial dietary overlap, with both species relying heavily on Po. desertorum; Munger (1984b) gives use of Po. desertorum as 23% of all identified prey items by Pm and 66% by Pc,  (28% and 8% of ant items respectively were not able to be identified by watching preying lizards).

         Does lizard size affect ant resource use?

        There is little comment in the literature on the relationship of prey size to horned lizard size.  Rissing (1981) found that juvenile P. platyrhinos ate proportionately fewer large ants and more small ants than adult individuals, but there was great variation among lizards. Suarez et al. (2000) found that adult P. coronatum  “predominately ate the largest ant species, juveniles...ate more smaller species”. However, this was not true for one of the four sites they studied. Amazingly, Whitford and Bryant (1979) report that hatchling P. cornutum "fed exclusively on Po. rugosus and Po. desertorum,"  both large species.

        In the present study there are apparent trends in the captures of ant species, both in terms of numbers and weight, as lizards increase in size (Tables 4, 6).  This must be due to interactions of lizard size, ant size, ant behavior (about which we know something) and lizard behavior towards ants (very little known).  With regard to numbers:

        (1) The small ants, I. pruinosum and C. insana, are numerically important only for the smallest lizards, but even then rank only 3rd and 4th.  Their IV decreases with increasing lizard size class, and is negligible for the largest Pc.

        (2) However, the small Ph. rugulosa-xerophila rank 1st numerically for 22-52 mm Pm and were sometimes taken in large numbers by medium and large Pc. This might seem like erratic instances due to small number of samples, however the samples that contained large numbers are from four different lizards in four years.  In five feedings the fraction of Ph. rug-xer  to total ants was: 422/426, 402/426, 116/184, 759/932; so, it seems to be a consistent phenomenon of the lizards’ feeding.  Feeding on Pheidole is made more rewarding because the species forage in columns up to at least 25 feet in length (Chew, personal observation) and a small number of majors are present in the columns and around nest entrances. The majors weigh much more than the workers (Table 1); they were present in 46% of the fecal pellets containing Pheidole.

        The above species of ants contrast markedly in their behavior. I. pruinosum and C. insana  move rapidly on the ground, often on recumbent plant stems in the heat of the day; they scatter away rapidly when disturbed.  Although they recruit in large numbers to honey-solution bait, there is limited occasion for such a large resource on the ground, eg. dead insects and vertebrates, and most of their foraging is as scattered individuals on the ground and up in many shrubs seeking plant exudates and insects. Therefore they must be a very limited resource for Phrynosoma. Ph. rug-xer forage in columns to scattered concentrated seed patches and move relatively slowly.  They are also concentrated at nest entrances as they return with seeds and do nest maintenance.  They are an opportunity for Phrynosoma to capture large numbers after a chance encounter, which may explain the exceptional cases just described.  Large lizards, having a longer feeding period, may have more such chances.  The importance of chance in lizard feedings is supported by Shaffer’s (1979) observation of “prey switching” by Pm.  On some occasions Pm took predominantly small ants, especially Pheidole, rather than Myrmecocystus . He speculated these were times of greater activity of small ants after rain.

        (3) C.larreae, the largest of the small-sized ants (except for X. spinosus , which was of little or no importance) increased in importance with lizard size (except for one erratic value, Table 4).  This species lives only in the lower stems and roots of creosote bushes (Buren 1986, Chew personal obsevation); it had a high density of 258 nests/h (Table 1).  It feeds on insects and honeydew and may “cultivate” homoptera and scale insects.  Foraging behavior has not been described, but it is reasonable to expect that it is centered around creosote bushes, and with the number of such shrubs, it is probably a high incidence resource for horned lizards.

        (4) The numbers of the medium sized Po. imberbiculus and M. depilis remained consistently high in all size classes of lizards, except the largest Pc. As the total number of ants in feedings increased (Table 5), their relative abundance was maintained. Their contrast of habits and behavior expose imberbiculus   to more frequent capture by Phrynosoma  than depilis . The seed-harvesting imberbiculus  occurs in many small colonies (137/ha) and individuals feed individually on the ground. Creighton (1956) characterized the species as “forages singly and at a slow and steady gait, extremely docile, make little effort to escape”. M. depilis had few large colonies (22.2/ha).  A group of 100-500 workers will emerge from the nest and go in all directions on the ground and up into shrubs after insects and plant exucates (Shaffer and Whitford 1981, Chew unpublished). When attacked, an individual either remains motionless, presumably making it difficult for a horned lizard to detect, or  rapidly runs away.

        (5) The largest ants, Po. desertorum and N. cockerelli were not taken by the smallest lizards. Possible these small individuals do not have tongues that are large enough or sticky enough to hold large ants, or that their head structures are not yet adequate for mastication of large ants (Montannuci d1989).  (The numerical importance ofdesertorum remained low until it became the dominant species of the largest Pc. Numbers of N. cockerelli were always low.  Capture of cockerelli is probably limited because it is diurnal only during cooler and moister days.

                 The interaction of numbers of items and their weight determines the importance of an ant species in terms of nutrition for a horned lizard. If, in Table 4, one sets an importance value by weight of 0.07 as the threshold IV for an ant species to be considered important nourishment, then:

        (6) The seven small ants are important in only six instances, all but one for 22-36 and 37-52 mm lizards.

        (7) Medium-sized Po. imberbiculus consistently ranks 1st in IV except for the largest Pc.  Medium sized M. depilis ranks 2nd for Pm but only 4th for small and small-medium Pc.

        (8) The large Po. desertorum  has very high weight IV only for the largest Pc although a few of these large ants added to the diet of some smaller lizards brings IV to slightly greater than 0.07.  The largest ant, N. cockerelli is important weight only for 53-65 mm Pc.

        The indices of food niche overlap (Table 9) summarize the effect of lizard size on resource use.  Particularly for Pc, overlap decreases as the size difference increases. For example, when there is a difference of one size class (n=3 cases) niche overlap on a weight basis averages 0.59; when there is a difference two size classes (n = 2) overlap averages 0.36; and in the one case of a difference of three size classes (22-36 mm compared with 66-103 mm) overlap is only 0.08. Overlap is markedly least in comparisons involving the largest size Pc.   Overlap for Pc on a numbers basis is about the same.  Overlap based on numbers for Pm declines in the  same way, but not to same degree; the least overlap is still 0.58; but for Pm there is no decline of overlap based on weight.

         Table 9 also summarizes diet overlap between species. Overlap of species is quite high within the 22-36 and 37-65 mm classes.  For 53-65 mm lizards, overlap in terms of numbers of ant species is low, 0.28 ( this seems erratic since overlap with 66-103 Pc  is 0.48),  but on a weight basis it is 0.53. The most extreme low overlaps are by the largest Pc with smaller lizards, both within and between species. For Pm and Pc in the same size class at this site there are some differences in resource use, which might be some help in their coexistence, but not at all clearly so.

         If Pm and Pc have an ability to gauge their predation so as to feed efficiently, as suggested by Munger (1984c), the relationship of lizard size x ant size is adaptive.  Small lizards are satiated by less weight intake than large ones (Table 7).  In the same time for feeding, a small lizard can “afford” to take small ants; 27 I. pruinosum equal 20% of the average total feeding of small Pc, whereas they are only 1.5% of the average feeding of the largest Pc.  In order to meet its food requirement, the largest Pc  must feed longer, or more rapidly, or learn to prey on larger ants. “Haste” in feeding may explain the greater ingestion of pebbles by the largest Pc, whose feedings  collectively contained 38 pebbles, whereas those of smaller size classes contained only 1-6.

        How do Phrynosoma move through their habitat?

        As shown in Table 8 the feedings of individual animals are not exceptionally similar. The overall average for individuals in all size classes and both species is similarity of 0.52 for numbers and 0.59 for weight.  This must reflect temporal and spatial variations in their range of activity within just a few days.  Comparisons between individuals, with spatial, temporal and even annual variations, have an overall average similarity of 0.36 for numbers and 0.35 for weight. Considering the extreme time intervals between the feedings of some individuals, one could expect much lower similarities.

        By watching individual lizards, Shaffer and Whitford (1981) and Munger (1984b) saw that Pm did most of its feeding on ants away from colony entrances.  Whitford and Bryant (1979) saw Pc also feeding more often away from colony entrances, whereas Munger (1984b) saw most feeding near nests.  Whitford and Bryant (1979) measured 73% of the feeding time of Pc  was from 0900 to 1100, and that individuals moved an average of 46.8 m/d with “random zigzag movements between feedings, rarely crossing their own trail”.  Shaffer (1979) measured activity of Pm  as greatest during “random walking and running” 0800-1000 and during the next hour under vegetation; “ants were rarely taken on [Myrmecocystus] nest disks...prey ...were most often encountered...in the shade, although any prey encountered in the open was also readily taken”.

        Munger (1984a) measured that Pm had relatively small “limited” home ranges: females 0.14 + 0.28 ha, males 0.41 + 0.96 ha, whereas Pc females moved over 1.38 + 1.22 ha and males 2.40 + 2.36 ha.  He estimated that these areas of activity were more limited than if the lizards moved about randomly, and that their areas overlapped less than if they had been located randomly.  He concluded that the large standard deviations meant that some lizards did not have a limited home range. In the present study the presence of Pm and Pc was noted most thoroughly in August 1958-July 1959 and July-August 1977.  In 1958-1959 34 different Pm and 2 Pc  were found in 3.344 ha, which is 6.1 and 0.36 lizards per ha respectively. In 1977, 31 Pm  and 15 Pc  were found in 3.344 ha, which is 9.3/h and 4.5/ha.  Only 10 Pm and 1 Pc were captured twice. Different individuals were found at locations very close together, at different times, so there was either an absence of limited ranges of  activity and movement of individuals through the site, and/or high mortality.  Munger (1986) measured a mortality rate of 0.52-0.86 per season for Pm and 0.14-0.56 for Pc.

        The foraging behavior of ants, individual versus group, must make an important difference in their predation by horned lizards. Davidson (1977b) showed experimentally, using arrays of concentrated and dispersed seeds, that group and individual foraging seed-feeding ants “deploy their workers in different spatial configurations”.  Although independently foraging species encountered concentrations of seeds and removed some, most of their forage was dispersed seeds.  Group foragers harvested predominantly the concentrated seeds.  Independent foragers spent more time searching for seeds and were active a greater percentage of the time than group foragers.  One can predict then that Ph. desertorum, Po. imberbiculus, Po. desertorum  and N. cockerelli  are more available as prey, on a per nest basis, than Ph. rugulosa  and Ph. xerophila .  For ants that are not seed feeders, much of their foraging is for nectar, honeydew and insects up in shrubs, and they recruit groups only when in the unlikely event of a sizeable resource. Hence these species have a low exposure to horned lizards.

        For heuristic value, consider a female Pm .  At the present site her area of activity of 0.14 ha (assumed from Munger 1984a) would contain an average of 198 ant colonies.  These colonies are probably randomly distributed interspecifically (Davidson 1977a, Chew unpublished).  Each of the colonies of independently foraging harvester ants: 19 Po. imberbiculus, 11 Po. desertorum, 10 Ph. desertorum and 1 N. cockerelli can be visualized as providing a “halo” of individuals dispersed during long periods of activity.  In contrast, the 53 colonies of group foraging harvesters, Ph. rug-xer provide short-term concentrated groups of individuals in foraging columns and around nest entrances.  The 36 colonies of C. larreae  probably provide concentrated individuals around creosote bushes.  Colonies of the other ants: 19 I. pruinosum, 46 C. Insana and 3 M. depilis probably provide variable dispersions of individuals as they feed on the ground and in shrubs.  I speculate that Pm moving haphazardly through such an array, feeding opportunistically on ants with different susceptibilities of capture would produce feedings such as sampled herein.

        It is possible that for Pc there may be a learning component as lizards grow through a longer life span than Pm such that they learn to seek and capture the large Po. desertorum , and ignore smaller ants or spend less time trying to capture them.

        These speculations are susceptible to modelling and the details of the feedings in this study are available to anyone so inclined, and to verification by much more intensive one-on-one observation of individual lizards.  I think the present study and earlier ones make it obvious that much more sampling is necessary before any firm conclusions can be drawn about food habits of species of Phrynosoma, and that there will be very much site specificity in feeding habits.

        Pm  and Pc as “ant specialists”.

        Phrynosomaspp. are usually described as “ant specialists”.  Numerically this is certainly true of Pm and Pc in this study.  However, other kinds of insects are present in most of their feedings (Table 10).  Often these other insects make an important contribution to the total weight of the feeding, an amount equal to 25-77% of the weight of ants in the feeding.  Most of the insects in feedings were ground dwelling beetles.  Although termites are an important food item of some horned lizards, the termites in the study site characteristically build soil tunnels to and around their feeding sites on dead shrubs and grass clumps, so their presence may be largely not perceived by Pm and Pc.
 
 

        LITERATURE CITED
 

Barbault, R., and M-E Maury. 1981. Ecological organization
    of a Chihuahua Desert lizard community.
    Oecologia  51:335-342.
Buren, W. F. 1986. A review of the species of Crematogaster,
    sensu stricto, in North America (Hymenoptera, Formicidae)
    Part II.  Descriptions of new species. J. Georgia
    Entomological Society 3:91-121.
Chew, R. M. 1977. Some ecological characteristics of the
    ants of a desert shrub community in southeastern Arizona.
    American Midland Naturalist 98:33-49.
Chew, R. M., and J. De Vita. 1980. Foraging characteristics of a desert ant
     assemblage: functional morphology and species separation.  J. Arid
     Environments 3:75-83.
Creighton, W. S. 1956. Studies on the North American representatives of
     Ephebomyrmex  (Hymenoptera: Formicidae). Psyche 63:54-66.
Davidson, D. W. 1977a. Species diversity and community organization in
    desert seed-  eating ants. Ecology 58:725-724.
_________________ 1977b. Foraging ecology and community organization in
    desert seed-   eating ants. Ecology 58:725-737.
Gaspar, C., and F. G. Werner. 1976. The ants of Arizona: An ecological
    study of ants  in the Sonoran Desert. US/IBP Desert Biome Research Memo.
    73-560. Utah State University, Logan. 14 pp.
Howard, C. W. 1974. Comparative reproductive ecology of horned lizards,
    (Genus Phrynosoma ) in southwestern United States and northern Mexico.
     J. Arizona Academy of Science 9:108-116.
Montanucci, R. B. 1989. The relationship of morphology to diet in the
    horned lizard   genus Phrynosoma . Herpetologica 45:208-216.
Munger, J. C. 1986a.  Home ranges of horned lizards (Phrynosoma ):
    circumscribed and exclusive?  Oecologia 62:351-360.
__________________ 1984b. Optimal foraging? Patch use by horned lizards
    (Iguanidae:Phrynosoma ). American Naturalist 123:654-680.
__________________ 1984c. Long-term yield from harvester ant colonies:
    implications for horned lizard foraging strategy.  Ecology 65:1077-1086.
__________________ 1986. Rate of death due to predation for two species
    of horned lizard, Phrynosomacornutum  and P. modestum .
    Copeia 1986:820-823.
Pianka, E. R. 1973 The structure of lizard communities. Annual Review
    Ecology  and Systematics. 4:53-57.
Pianka, E. R., and W. S. Parker. 1975. Ecology of horned lizards: a
    review with  special reference to Phrynosomaplatyrhinos .
    Copeia 1975: 141-162.
Rissing, S. W. 1981. Prey preferences in the desert horned lizard:
    influence of prey  foraging method and aggressive behavior.
    Ecology 62: 1031-1040.
Shaffer, D. T. 1979. Behavioral responses of a predator, the
    round-tailed horned lizard (Phrynosomamodesteum ) and its prey, honey-pot
    ants (Myrmecocystus spp.). MS in Biology Thesis, New Mexico State
    University, Las Cruces, 25 pp.
Shaffer, D. T., and W. G. Whitford. 1981. Behavioral responses of a
    predator, the   round-tailed horned lizard, Phrynosoma modestum , and
    its prey, honey pot   ants, Myrmecocystusspp.
    American Midland Naturalist 105:209-216.
Suarez, R. and  T. J. Case. 2000. Prey selection in horned lizards following the
    invasion of Argentine ants in southern California.
    Ecological Applications   10:711-725.
Whitford, W. G., and M. Bryant. 1979. Behavior of a predator and its prey: the
    horned lizard (Phrynosoma cornutum ) and harvester ants
    (Pogonomyrmex  spp.). Ecology 60:686-694.

 

Insects other than ants P. modestum P. cornutum Size Class n % total diet n % total diet 22-36 mm 2/7 77.4, ? 2/3 5.5, 16.8 37-52 mm 7/8 11.4, 15.5, 25.9, 30.1, 39.9, 43.3, 56.6 3/4 16.3, 17.0, 25.5 53-65 mm 7/8 5.5, 5.8, 20.5,  35.5, 36.6, ?, ? 3/4 11.0, ?, ? 66-108 mm     7/7 3.5, 7.2, 20.6, 27.1, 28.8, 50.2, 63