Character displacement in some Cnemidophorus lizards revisited: A phylogenetic analysis

Abstract

Ecological studies have demonstrated the role of competition in structuring communities; however, the importance of competition as a vehicle for evolution by natural selection and speciation remains unresolved. Study systems of insular faunas have provided several well known cases where ecological character displacement, coevolution of competitors leading to increased morphological separation, is thought to have occurred (e.g., anoline lizards and geospizine finches). Whiptail lizards (genus Cnemidophorus) from the islands of the Sea of Cortez and the surrounding mainland demonstrate a biogeographic pattern of morphological variation suggestive of character displacement. Two species of Cnemidophorus occur on the Baja peninsula, one relatively large (Cnemidophorus tigris) and one smaller (Cnemidophorus hyperythrus). Oceanic islands in the Sea of Cortez contain only single species, five of six having sizes intermediate to both species found on the Baja peninsula. On mainland Mexico C. hyperythrus is absent, whereas C. tigris is the smaller species in whiptail guilds. Here we construct a phylogeny using nucleotide sequences of the cytochrome b gene to infer the evolutionary history of body size change and historical patterns of colonization in the Cnemidophorus system. The phylogenetic analysis indicates that (i) oceanic islands have been founded at least five times from mainland sources by relatives of either C. tigris or C. hyperythrus, (ii) there have been two separate instances of character relaxation on oceanic islands for C. tigris, and (iii) there has been colonization of the oceanic island Cerralvo with retention of ancestral size for Cnemidophorus ceralbensis, a relative of C. hyperythrus. Finally, the phylogenetic analysis reveals potential cryptic species within mainland populations of C. tigris.

Ecological studies have repeatedly shown the central role competition plays in structuring communities (1–3), but the importance of competition as a vehicle for evolution by natural selection and speciation remains unresolved (4–6). The joint evolution of morphological character differences as a result of competition between species has come to be known as ecological character displacement (7, 8). By this model, the phenotypic changes that occur as a result of character displacement represent adaptations that allow optimal resource utilization. Character displacement is difficult to unambiguously confirm. The ideal study system would possess replicated populations varying only in the presence or absence of particular species. Few studies have gone beyond correlating morphological character-states with the presence or absence of potential competitors in attempting to meet the criteria for establishing that character displacement has occurred (8–10). Strong experimental evidence has been provided that stickleback (genus Gasterosteus) populations established when the sea level dropped in the late Pleistocene have undergone subsequent character displacement (6, 9, 11).

The whiptail lizard (Cnemidophorus) system on islands of the Sea of Cortez and the surrounding mainland is another case where character displacement has been implicated (12, 13). Members of the lizard genus Cnemidophorus (family Teiidae) are diurnal, insectivorous lizards exhibiting a “widely foraging” hunting strategy (14). Two species (Cnemidophorus tigris and Cnemidophorus hyperythrus) and related species and subspecies commonly occur in Baja California and on islands in the Sea of Cortez (15–17) (see Fig. 1). The two are broadly sympatric throughout Baja California with C. tigris always being larger than C. hyperythrus (12, 13).

Figure 1.

Location of C. hyperythrus and C. tigris, along with related island endemics, sampled around the Sea of Cortez. The hatched area in the northern gulf area delineates the subset of the range of C. tigris where it occurs allopatrically. In the rest of the area covered by the map, C. tigris is sympatric with C. hyperythrus (across the Baja California peninsula), C. burti (Arizona and Sonora) and other larger Cnemidophorus species elsewhere in coastal Mexico. Small populations of C. labialis (sexlineatus species group) occur along the Pacific coast of the Baja California Norte. This species is similar in size to C. hyperythrus, but the two species are not syntopic (13). Numbers and lowercase letters refer to map labels; abbreviations following locations are used in the text. BCN, Baja California North; BCS, Baja California South; and BCC, Baja California Cape region. Numbers following locations refer to sample size if multiple sequences (full plus partial) were obtained. C. hyperythrus collections are indicated by an alphabetic symbol: a, San Diego, CA (SD); b, Bahia de los Angeles, BCN (BLA) (2); c, El Arco, BCS (EA); d, San Ignacio, BCS (SI); e, La Paz, BCC (LP); f, I. San Marcos (SM) (1 + 2); g, I. Coronados (CO); h, I. Carmen (CA) (2 + 2); i, I. Monserrate (MO) (1 + 2); j, I. San Jose (SJ) (1 + 2); k, I. San Francisco (SF) (1 + 1); l, I. Espiritu Santo (ES) (1 + 2); m, I. Cerralvo (CE) (1 + 3). Cnemidophorus tigris collections are indicated by a number: 1, Ortiz, Sonora (OR); 2, Isla Tiburon (TI) (2 + 2); 3, Punta Penasco, Sonora, Mexico (PP) (3); 4, Catavina, BCN, (CT) (1 + 1); 5, Bahia de los Angeles BCN (BLA) (5 + 2); 6, I. Smith (SMI) (2); 7, I. Angel de la Guarda (ANG) (1 + 2); 8, I. Partida Norte (PN) (4); 9, I. San Esteban (SE) (1 + 2); 10, I. Salsipuedes (SL) (4); 11, I. San Lorenzo Norte (SLN) (3); 12, I. San Pedro Martir (SPM); 13, I. Cedros (CD); 14, El Arco, BCS (EA); 15, San Ignacio, BCS (SI); 16, Punta Abreojos, BCS (PA); 17, I. San Marcos (SM) (1 + 1); 18, I. Carmen (CA); 19, I. Santa Catalina (SC); 20, La Presa de San Pedro, BCS (SP); 21, I. San Francisco (SF); 22, I. Espiritu Santo (ES) (2); 23, La Paz, BCC (LP). Islands within the 100-m depth contour were connected to the Baja peninsula during the last 14,000 years (Cedros and San Marcos are land-bridge islands as well).

C. hyperythrus remains fairly constant in size [≈62 mm upper decile of snout-to-vent length (SVL)] whereas C. tigris varies from 130 mm SVL in southern Cape region to 92 mm SVL in northeastern Baja. Only C. tigris lineages, C. hyperythrus lineages, or both occur on the gulf islands. C. tigris is allopatric in the region surrounding the northern gulf. From southwestern Arizona to southern Mexico C. tigris is not syntopic with C. hyperythrus but instead is found syntopic with either the larger Cnemidophorus burti or Cnemidophorus costatus.

The result of the mainland character shifts is that the Sinaloa, Mexico, C. tigris is as small in size as C. hyperythrus at the same latitude, climate, and vegetation type across the Sea of Cortez in the Cape region of Baja California. It has been hypothesized that the process of ecological character displacement is the driving force for these body size changes (12, 13).

Case (12) combined niche overlap and relative abundance data to predict optimal body sizes for one- and two-species whiptail guilds. The predictions for solitary C. tigris held on five of six oceanic islands (older than 1 million years) in the gulf. The predictions also held for the two species Baja guild. C. tigris found alone on recently separated land-bridge islands show no significant differences in body-size relative to the nearest mainland population (no oceanic islands contain both species).

All populations of C. hyperythrus are small relative to Baja California C. tigris. The oceanic island of Cerralvo contains a solitary species, Cnemidophorus ceralbensis, thought to be closely related to C. hyperythrus. C. ceralbensis has a body size near the solitary species body size expected ecologically for single-species islands. The only other solitary insular C. hyperythrus occurs on the island Monserrate and is significantly larger than individuals from the Baja peninsula.

The analysis of Case (12) addressed the question “how many island populations have body sizes for the two species which were consistent with character displacement.” However, without an understanding of the ancestral–descendent relationship between populations, these size differences between extant taxa could not be translated into arguments about evolutionary size changes; size differences may be due to the assortment of different sized clades on islands rather than in situ body size evolution. Case and Sidell (18) referred to these alternatives as size adjustment vs. size assortment. Therefore, using a phylogenetic reconstruction based upon nucleotide sequences of the cytochrome b gene, we ask the following.

(i) Where do inferred body size changes occur along the phylogeny for the two species? Using an independent contrasts approach, are body size contrasts significantly larger on nodes with an inferred change in the geographic sympatry/allopatry condition relative to those without?

(ii) How many of these inferred body size changes occur coincidentally with an inferred change in the geographic sympatry/allopatry condition with congeners, and of these cases, how many are in the direction predicted by the character displacement hypothesis? How many size changes occur in a direction opposite to predictions?

(iii) Conversely, how many inferred shifts in the geographical allopatry/sympatry condition are accompanied by inferred body size changes in the direction predicted from the character displacement hypothesis? How many such biogeographic shifts are not accompanied by the predicted body size change or the inferred body size change is in the opposite direction from a character displacement prediction?

Size assortment can arise as the result of the greater persistence and invasion resistance of some species combinations based on their size compatibility (18–20). No local size evolution need be invoked although some size variation is necessary to account for the range of sizes in invaders. The populations begin preadapted to a particular competitive regime.

We quantify the degree of size assortment as follows. Under the size assortment hypothesis, species in a clade of closely related species with a distinctive and similar body size should all hold the same sympatry/allopatry condition. For example, taxa in a clade comprised solely of C. tigris with intermediate relative body size, by the size assortment hypothesis, should occur solely on one-species islands. A clade of several populations of C. tigris, all with a large body size, should all occur on two-species islands. The number of taxa (or isolated populations) that conform to this expectation relative to the number that violate it, gives a measure of the strength of size assortment in shaping Cnemidophorus body size distributions.

MATERIALS AND METHODS

DNA Sources and Extraction.

The species and localities within species used in the phylogenetic analysis are listed in the legend to Fig. 1. For many locations or species full or partial sequences were obtained from multiple individuals to ensure that observed variation between locations was not due to inadequate sampling within location. Extraction of DNA follows described methods (21).

PCR Amplification and Sequencing.

The phylogenetic analysis was based upon 887 bp (excluding primers) of mtDNA sequence. The targeted fragment was initially amplified using the primer 5′-GCASCAWAAAARGGAGA-3′ and H15560 (22). The former primer was designed by aligning human mtDNA to Cnemidophorus uniparens (23) sequences. Initial results from cloned PCR products uncovered a partial duplication of cytochrome b containing stop codons. The duplication was 5′ to the actual gene. To avoid amplifying the duplication a primer was designed for the light strand from the distal end of the psuedogene. Subsequent amplifications used the 5′-TTYGTTGTTTTGMGGT-3′ in conjunction with H15560. Two methods were used to prepare PCR products for sequencing: (i) PCR products were cloned into pCR II vector (TA Cloning Kit; Invitrogen) following manufacturer’s instructions. Chemically denatured plasmid DNA was sequenced using the chain termination method (24); and (ii) single-stranded PCR products were produced using the Dynol solid phase sequencing system, followed by sequencing.

Phylogenetic Analysis.

Phylogenetic relationships were inferred by adopting a maximum parsimony criterion (25) with the branch and bound algorithm (26). Each nucleotide position was considered a unique character. The effects of differential weighting of transitions vs. transversions was evaluated by measuring the level of skewness, as measured by g1 (27), in tree-length distributions of randomly constructed trees. In the present case an equal weighting of transitions vs. transversions produced the most negative g1 values; therefore all changes were given equal weights.

The taxa used in phylogenetic reconstruction contain representatives from three Cnemidophorus species groups: C. tigris, Cnemidophorus deppii (containing C. hyperythrus and C. ceralbensis), and Cnemidophorus sexlineatus. These groups are monophyletic (excluding parthenogenetic forms) based upon morphologic and karyotypic characters (15, 16). The phylogenetic relationships of all taxa were inferred simultaneously. Land-bridge island taxa were included to improve phylogenetic resolution. Confidence was evaluated using the bootstrap technique (28). Alternative topologies were tested using a maximum likelihood algorithm (29).

Body size (SVL) was optimized as a continuous character (30) on the most parsimonious topology. The upper decile was used rather than mean to avoid influences of age distribution (31). The body-size reconstruction was conducted separately for the C. tigris group and the C. deppii group. Land-bridge island taxa were pruned prior to size reconstruction.

The land-bridge island taxa were excluded for several reasons. First, we must infer the historical allopatry/sympatry condition based on the present biogeographical ranges of these species because the fossil record in this region is not detailed enough (32, 33). We assume, for example, that islands which now have a single species of Cnemidophorus have always lacked the other species. This assumption is probably not correct for Holocene land-bridge islands, because they undoubtedly began their recent island existence 14,000 to 6,000 years ago with several more species than they have today (12, 34, 35). We focus on oceanic islands, and mainland comparisons where the Gulf of California has been a barrier to dispersal. Second, observed body size distributions on land-bridge islands are likely to be nonequilibrium because of changing species compositions leading to inaccurate estimates of ancestral states.

Two parsimony algorithms are available for optimizing the evolution of continuous character-states on phylogenies, squared-change (30) and linear parsimony (36). Squared change algorithms seek to minimize the amount of squared change along each branch across the entire tree simultaneously (37); the absolute amount of evolution is not necessarily minimized, and some degree of change is forced along most branches. Linear parsimony algorithms seek to minimize the total amount of evolution and consider only the three nearest nodes when calculating the final ancestral character states. The result of this local optimization is that changes are inferred on very few or single branches.

The imprint of character displacement in a phylogenetic analysis will appear as single change along the branch once the competitive environment has changed. For example, natural selection for character release cannot occur until one species is released from the ecological constraints of its competitor. Squared-change parsimony could potentially spread the change that occurred on a single branch across two or more branches. Squared-change parsimony could also blur changes in the opposite direction of ecological predictions. Therefore, we used the linear parsimony approach because it provides a reconstruction that provides better resolution of changes concentrated on a single branches. However, the qualitative results we discuss do not change using squared-change parsimony.

Independent contrasts were calculated using two methods. First the commonly used method proposed by Felsenstein (38) was employed. In addition, a recently proposed method that incorporates branch-length information was also used (39). Values were compared using a nonpaired Student’s t test. Because of unequal variances contrasts were log-transformed prior calculating the t value.

RESULTS

Phylogenetic Reconstruction.

The sequence analysis provided a total of 38 sequences. Divergence varied from 0.007%, between mainland C. hyperythrus populations (SD and EA), to 16.9% between C. tigris from SLN and C. uniparens. All cases of within population variation were <1%.

The results of the phylogenetic and bootstrap analysis are shown in Fig. 2. The two equally most parsimonious trees are identical except for alternative trees found for closely related populations within C. hyperythrus. These alternatives are depicted as a polytomy using a strict consensus rule (Fig. 2a). Branches are drawn proportional to the inferred number of nucleotide substitutions under a parsimony criterion.

Figure 2.

Phylogenetic relationships of Cnemidophorus species. The tree contains the two groups: “tigris” and “deppii.” For taxa described as either C. hyperythrus or C. tigris, the locations abbreviations used in Fig. 1 are used as labels; for all others the species designation is used. Taxa from single species oceanic islands are shown in bold. Taxa from land-bridge islands are shown as open rectangles. (a) The most parsimonious tree with branches drawn proportional to the inferred number of nucleotide substitutions (d). The linear parsimony reconstruction (after pruning land-bridge islands) of upper decile of SVL (mm) is shown. Changes of 5% or greater are shown in gray. Evolutionary ecological events are given roman numerals corresponding to Table 1. (b) Consensus tree based upon a bootstrap analysis (400 replications). Nodes with values of ≥50% are shown.

The bootstrap consensus tree (Fig. 2b) shows that fairly strong support exists for the majority of nodes across the phylogeny and highlights several features. First, both the C. tigris and C. deppii groups are monophyletic in the majority of bootstrap replications. Second, five well-supported and distantly related clades are observed within the C. tigris group (i) northern Sonora–northern Baja plus the four oceanic islands SE, SPM, SLN, SL; (ii) I. Santa Catalina; (iii) mid-Baja plus associated land-bridge islands (iv) southern Baja plus the oceanic islands of ANG and PN; (v) Cape region C. t. maximus subspecies.

Body Size Reconstruction.

Also shown in Fig. 2a is the result of the linear parsimony reconstruction of body size evolution and the inferred changes of allopatry/sympatry. Observed body sizes for each taxon are shown at the tips and inferred values are placed next to each node. Several equally most parsimonious reconstructions are possible. Only two instances of an ambiguous reconstruction occurred; both were a difference of 0.5 mm and the higher value was used. Inferred changes of competitive environment are depicted by roman numerals corresponding to Table 1. Two decreases of body size on single species oceanic islands are inferred: (i) a decrease from 88 to 79 mm for the node basal to the monophyletic group SE, SPM, SLN, SL; and (ii) a decrease from 88 to 81.8 mm for C. tigris on the single species oceanic island SC.

Table 1.

Evolutionary and ecological events mapped onto the phylogeny in Fig. 2

Evolutionary ecological event	Change of community structure	Direction or magnitude of size change consistent with most recent change of competitive environment?
C. tigris
I. San Esteban clade	Sympatry → allopatry	Yes
II. Salsipuedes	No	Yes
III. Base of northern Sonora	Sympatry → allopatry	Yes
IV. Northern Baja	Allopatry → sympatry	No inferred size change
V. Santa Catalina	Sympatry → allopatry	Yes
VI. Base of southern Baja	Relative body size reversal	Yes
VII. Angel/Partida	Sympatry → allopatry	No inferred size change
VIII. C. t. maximus	No	Yes
C. hyperythrus
IX. Base of C. hyperythrus clade	Allopatry → sympatry	Yes
X. Monseratte	Sympatry → allopatry	No inferred size change
XI. Base of Baja clade	No	Yes

The changes were inferred after pruning all land-bridge islands. The analysis assumes a substantial body size change to be 5% and the ancestral body size for C. tigris was 88 mm or less. 

Independent contrast values (39, 40) were calculated using the inferred ancestral body sizes. The values were compared under a null hypothesis that no difference exists between nodes experiencing a change in allopatry/sympatry condition relative to those experiencing no such change. The null hypothesis was rejected for both the punctuated (t = −3.036, P = 0.0032) and the branch-length method (t = −2.289, P = 0.0168), with “change” nodes significantly larger than “nonchange” nodes.

Although the contrast values are significantly different, it is still possible changes occurred in the direction opposite to that predicted by size-mediated competition. Table 1 provides a synopsis of evolutionary and ecological events mapped onto the observed phylogeny. Table 1 shows that five of eight changes in allopatry/sympatry condition lead to an inferred change in body size consistent with predictions of ecological character displacement. In the remaining three cases no change occurred (no changes in the wrong direction). Table 1 also shows three cases where body size changed in the absence of a change in competitive environment. The direction of change in these three cases is consistent with the most recent change of allopatry/sympatry condition. No changes occurred in the direction opposite to that predicted by character displacement.

The role of size assortment in ecological communities where no change of relative body size was inferred since colonization is addressed in Table 2. The table includes both oceanic and land-bridge islands. Land-bridge islands are included because the process of size assortment can occur over an ecological time scale. Among oceanic islands (all single species) two of four lineages (SPM, SLN, SL, and CE) and four of seven populations are consistent with predictions of size assortment. The exceptions were large C. tigris on ANG and PA, and small C. hyperythrus on MO. For land-bridge islands all two species islands were consistent (6) and all single species islands (6) were inconsistent with predictions.

Table 2.

Island populations experiencing no relative size change and their concordance with size mediated competition

Islands experiencing no relative size change	Species/relative size	Consistent with size mediated competition?
Oceanic (all single species)*
San Pedro Martir	C. tigris intermediate	Yes
San Lorenzo Norte	C. tigris intermediate	Yes
Salsipuedes	C. tigris intermediate	Yes
Angel de la Guarda	C. tigris large	No
Partida Norte	C. tigris large	No
Cerralvo (C. ceralbensis)†	Intermediate	Yes
Monserrate	Small/intermediate	No
Land-bridge
Two species islands
San Marcos	All C. tigris large, C. hyperythrus small	Yes
Carmen	All C. tigris large, C. hyperythrus small	Yes
Espiritu Santo	All C. tigris large, C. hyperythrus small	Yes
San Francisco	All C. tigris large, C. hyperythrus small	Yes
San Jose	All C. tigris large, C. hyperythrus small	Yes
Coronado	All C. tigris large, C. hyperythrus small	Yes
Single species islands
Smith	All C. tigris large	No
Danzante	All C. tigris large	No
Tiburon	All C. tigris large	No
Cedros	All C. tigris large	No
Mejia	All C. tigris large	No
Pond	All C. tigris large	No

^*Brackets indicate a single evolutionary size change leading to the present size in the denoted monophyletic taxa. Size changes were not inferred for land-bridge taxa (see text). 

^†Although C. ceralbensis has likely changed competitive environment relative to its previous state in Sonora, it is presently intermediate in size between the C. hyperythrus and C. tigris on the adjacent mainland. No inferred body size change has occurred thus size assortment.  

Testing Alternative Phylogenies that Indicate Minimum Size Evolution.

The minimal amount of evolution expected from size assortment can be used to motivate alternative phylogenetic hypotheses (Fig. 3) to those depicted in Fig. 1. Minimum Size evolution hypotheses I and II depict all C. tigris on oceanic islands as monophyletic—i.e., small size evolved only once and then these small ancestors invaded multiple islands. Two alternatives of minimum evolution are possible because the ancestral body size for all C. tigris could be small with a subsequent increase in size or large with a subsequent decrease. Minimum evolution hypothesis III shows all small C. hyperythrus monophyletic relative to the rest of the C. deppii species group. The topology generating the least amount of change necessary to explain the distribution of body sizes across the C. deppii group has C. ceralbensis sister to all C. hyperythrus taxa.

Figure 3.

General alternative phylogenies that would lead to the minimum amount of body size evolution for Cnemidophorus populations found on oceanic islands. The results of maximum likelihood tests are shown below the alternatives.

The input trees used for comparison with the two most parsimonious trees were generated by performing three separate parsimony analyses to find the shortest trees given the constraints set out in Hypotheses I–III. The results of the maximum likelihood (29) comparisons show that the minimum evolution hypotheses I and II are significantly worse than the most parsimonious trees, indicating that two separate instances of size change in dwarf C. tigris are necessary to explain the evolution of body size in the five oceanic island populations of C. tigris. In contrast the alternative topologies containing monophyletic clades of C. hyperythrus are not significantly worse than the most parsimonious trees. Therefore the simplest explanation remains that C. ceralbensis has retained its ancestral body size with a single reduction of body size inferred for all C. hyperythrus of the Baja peninsula.

DISCUSSION

Studies of insular faunas have yielded much of the evidence of character displacement. Examples include the anoline lizards of the Caribbean (41–43), mustelids of the British Isles (44) and geospizine finches of the Galápagos (8, 45). The phylogeny, in conjunction with the associated reconstruction of ancestral body sizes, allows us to draw the following conclusions about the evolutionary ecology of insular Cnemidophorus.

Evolutionary Change and Size Assortment Have Generated Much of the Observed Body Size Variation on Oceanic Islands of the Sea of Cortez and the Surrounding Mainland.

Case (12) used a coevolutionary model of resource partitioning to predict the optimal body size of Cnemidophorus populations found in one and two species guilds. The body size distributions of single species guilds were expected to fall between the distributions of two species guilds and this was found in most cases.

The body size reconstruction indicates that an evolutionary change of size consistent with general coevolutionary models of resource utilization has occurred at least twice on oceanic islands and also twice on the surrounding mainland. The phylogeny indicates that the single species island populations of SC and those in the SE clade have undergone size reductions in the absence of competitors. In the southern Baja region the phylogeny suggests a near simultaneous invasion of ancestral C. tigris and C. hyperythrus followed by character displacement of both species: an increase of C. tigris and a decrease in C. hyperythrus. Overall 8 of 11 evolutionary ecological events shown in Table 1 are consistent with size-mediated competition.

The analysis of size assortment shown in Table 2 provides support for size assortment in certain situations. First, an evolutionary adjustment of C. tigris on SE was followed by colonization of SPM, SLN, and SL. The colonization could be a result of over water colonization, in a stepping-stone manner, or vicariance, through tectonic break-up of a once consolidated land mass. Second, in the case of C. ceralbensis of Cerralvo, ancestral size has been retained since the transgulfian vicariant event the led to the evolution of this species and C. hyperythrus on the Baja peninsula. Finally, all two species land-bridge islands (6) contain large C. tigris and small C. hyperythrus.

If ecological rules involving resource competition are molding body size throughout the region, a more appropriate question now becomes, how are exceptions explained? Three general exceptions emerge from the present evolutionary analysis: (i) large C. tigris on the oceanic islands ANG and PA, (ii) small C. hyperythrus on the single species oceanic island MO, and (iii) large C. tigris on six single species land-bridge islands. No changes occurred in the direction opposite to that predicted by character displacement.

Populations with body sizes not conforming to the ecological expectations must have experienced different histories or environmental circumstance. One possible explanation for the lack of change on ANG, PA, and MO may be differences in resource availability (12, 13, 46). One result of the earlier ecological study by Case (12) was that the body size of the dwarf solitary C. tigris were smaller than optimal predictions. Prey availability was found to be lower on islands containing the dwarf C. tigris when compared with areas containing larger C. tigris including ANG/PN. Reduced prey availability may be due to over-exploitation or lower productivity. For the large island of ANG over-exploitation may not be possible. Increased prey availability through higher productivity may be responsible for the increased size of C. t. maximus in the subtropical southern cape region of Baja and also the slightly larger size of C. ceralbensis relative to the dwarf C. tigris.

The hypothesis that there has not been sufficient time for evolution to occur can be rejected for ANG/PA and MO because the branch lengths leading to their nearest mainland ancestors are longer than observed in other oceanic islands where change has occurred. A time argument can be invoked in the case of large C. tigris on single species land-bridge islands. Whether or not a second species was never present, or has recently gone extinct, the time since a change in sympatry/allopatry condition may have been too short for an evolutionary response.

Cryptic Species May Be Present on the Baja Peninsula.

The phylogeny also reveals moderate to very deep evolutionary splits between physically close Baja populations of C. tigris. For example SP and LP (see map, Fig. 1), separated by a physical distance of only ≈125 km, are two distinct lineages with over 10% sequence divergence. An even more remarkable example is the relationship between EA and BLA. While these two locations are located within 100 km of each other, the inferred phylogenetic relationship between EA and BLA is one of the most distant in the C. tigris clade. These relationships are not likely due to isolation by distance because the phylogeny also shows that some widely separated populations (e.g., SI-SP) are very closely related.

The distinctness of these lineages over short spatial scales may represent cryptic species generated through historical oceanic barriers. Historical existence of barriers, as supported by previous phylogenetic evidence (47–49), were between (i) the cape region and southern Baja, (ii) southern Baja and mid-Baja Vizcaino, and (iii) mid-Baja Vizcaino and northern Baja. The low elevational topography in these regions and the presence of Pliocene and Pleistocene marine terraces suggests that several areas were submerged during some intervals in these periods (50, 51). Other abrupt zones of contact between C. tigris lineages in other mainland areas are not unknown (52–54). Fine-scale molecular analysis should establish whether gene flow levels across these “hybrid zones” are low enough to consider these lineages separate species (55).

Competition Experiments Demonstrating Natural Selection Are Still Necessary.

The results of the historical analysis above are consistent with the previous ecological study which concluded that the presence or absence of competitors is influencing body size. However, proving that observed variation in body size are adaptations requires experimental manipulations to show that fitness is dependent on the abundance and frequencies of different sized competitors (9–11, 56). Divergent natural selection has been measured in Darwin’s finches (Geospiza) (57) and the threespine sticklebacks (Gasterosteus) (11). The effects of competition have been measured in field manipulations of reptile and amphibian systems (2).

Introductions of Anolis lizards to previously unoccupied islands of the Bahamas have yielded compelling results of the ecological and evolutionary consequence of competition (58, 59). Introductions and long-term field experiments may also be feasible on small land-bridge islands of the Sea of Cortez. Finally, the hybrid zones between the potential cryptic species in the C. tigris group may afford ongoing natural competition experiments.

ABBREVIATION

SVL

snout-to-vent length