Diversity and distribution of Odonata ( Insecta ) larvae along an altitudinal gradient in Coalcomán mountains , Michoacán , Mexico

Evaluating components of landscape diversity is essential for the implementation of efficient conservation strategies. We evaluated the diversity of Odonata larval assemblages from the Coalcomán mountains (CM), Michoacán, Mexico, and related it to local (site-level) habitat variables. Larvae were collected from shores, riffles and pools in five streams, counted and identified to species, twice per season during 2005. The Shannon Diversity Index (H’), Margalef’s Richness Index (R), Simpson’s Index as a dominance measure (D) and Pielou’s Equitability (J) were used to describe the assemblages, and Renyi’s Diversity Profiles were used to order diversity. A Bray-Curtis Similarity Index (BC) was used to evaluate beta diversity. Theoretical richness was estimated using non-parametric and parametric methods. A Canonical Correspondence Analysis (CCA) was applied to explore the relationships of species with site-level environmental variables. A total of 12 245 larvae from 75 species, 28 genera and 8 families were recorded. Over all sites, the dominant species were Erpetogomphus elaps, Macrothemis pseudimitans and Argia pulla. The number of species per locality ranged from 18 to 36, and a high number of species (76%) occurred with relative abundances lower than 1%. A differential distribution of species and abundance in streams, time and strata was observed. Renyi’s diversity profiles showed diversity was higher in spring and on shores. Most BC similarity values were smaller than 25%, indicating a high turnover rate in the CM. The high turnover rate reflects a differential distribution of the species along the altitudinal gradient, supporting the hypothesis of Mexico as a betadiverse country. According to the slope of Clench’s curve, a reliable list of species was gathered. The CM larval assemblage is currently the largest reported for Mexico, and our results support previous proposals of the CM as a species-rich area for conservation. Rev. Biol. Trop. 59 (4): 1559-1577. Epub 2011 December 01.

Given its topography, climatic diversity and geographical situation between two biogeographic regions (Neotropical and Nearctic), Mexico is a megadiverse country (Mittermeier et al. 1998).It is well known that the rate of species lost worldwide (Fraser et al. 1993) surpasses that by which taxonomic knowledge is increased.Thus, it is necessary to intensify studies focusing on diversity and from areas with an acceptable level of conservation.
The objective of this study was to assess the variation of diversity in Odonata larval assemblages along an elevation gradient and their relation with stream habitat variables by season and habitat.Local and landscape information on assemblage diversity may be useful in supporting conservation efforts and in following short and long term changes in their diversity in the Coalcomán Mountains.

Study area:
The Coalcomán Mountain Range (CM) in Michoacán,Mexico, The climate is predominantly tropical (annual average temperature 26.5°C), with a rainy period occurring in summer, and average annual precipitation ranging from 700mm in the lowlands to 1 300mm in the highlands (Antaramián 2005).The CM is one of the least disturbed regions in Michoacán.The studied assemblages are located in streams within Hydrological Priority Region 26 (Río Coalcomán and Río Nexpa; Arriaga et al. 1998), and Terrestrial Priority Region 115 (Sierra de Coalcomán; Arriaga et al. 2000) in Michoacán State, Mexico.The sampled streams included Ticuiz (TZ), Estanzuela (EZ), Río Pinolapa (RP), Colorín (CL) and Chichihua (CH).Details on water channels and averages of physicochemical data for the streams have been published by Novelo-Gutiérrez & Gómez-Anaya (2009), and are summarized in Table 1.
Larval collection and physicochemical measures: Larval sampling was conducted twice per season during 2005, as follows: February 17-21(winter); March 29-April 2 and May 31-June 4 (spring); July 6-10 and August 17-21 (summer); September 28-October 2 and November 8-12 (autumm) and January 18-22 (2006, winter).Larvae were collected during one day per stream, along a 500m stream segment using a D-frame aquatic net (0.2mm mesh).A stratified random sampling design was applied for stream margins, riffles and pools.A total of 12 samples were taken each time (3 samples in riffles, 3 samples in pools and the remaining 6 samples along margins).All samples were preserved in 96% ethanol with a replacement before 24h had elapsed.Odonata larvae were separated from the other fauna and stream debris using a stereomicroscope, counted and identified to species.All specimens were deposited in the Instituto de Ecología, A.C., Entomological Collection (IEXA).Physicochemical data (pH, temperature, dissolved oxygen, and conductivity) were recorded during each trip using a digital water analyzer (ICM Model 51500, Industrial Chemical Measurement, Hillsboro, Oregon, USA).Stream width, depth and current velocity were measured using a metric power tape and a digital flow meter (Global Water Flow Probe, Forestry Suppliers, Jackson, Mississippi, USA).At each point where we measured the stream's width, several measures of water depth and velocity were taken and averaged to calculate discharge as: width x average depth x average current velocity.Finally, gradient was calculated according to Resh et al. (1996).
For data analysis a one-way analysis of variance (ANOVA) was used to analyze the effect of stream on gradient, and a two-way multivariate analysis of variance (MANOVA) was applied to analyze the effect of stream and season on physicochemical variables.If significant effects were detected, we performed multiple comparisons using the Bonferroni t-test.All tests were performed using STATISTICA (StatSoft 2006).

Diversity indices:
Richness and composition, Shannon Diversity Index (H'), Margalef's Richness Index (R), Simpson's Diversity Index as a dominance measure (D), and Pielou's Equitability (J) (Moreno 2001) were used to describe the assemblages.To compare the diversity of the streams, Renyi's profiles were generated (Tóthmérész 1995, 1998, Southwood & Henderson 2000, Jakab et al. 2002) because they include the logarithm of species number, H', D, and the logarithm of Berger-Parker diversity (Tóthmérész 1995).The profiles also have a scale parameter (α) such that when the value of the parameter is low, the method is extremely sensitive to the presence of rare species, but as the value of the parameter increases, the profile becomes less sensitive.As a large scale parameter, the method is sensitive only to the more frequent species.The result of this scale-dependent characterization of diversity can be presented in graphic form to visualize relationships among communities: the community's diversity profile'.If the curves of two diversity profiles intersect, then one of the communities is more diverse for rare species, while the other is more diverse for frequent species.
Gamma richness estimation: An estimation of the theoretical richness using nonparametric methods (presence/absence: Chao2, Jack2, Bootstrap; abundance: ACE, Chao1) was carried out using EstimateS 8.0 (Colwell 2006).Additionally, parametric methods (richness estimators that use the observed species accumulation curve for modeling the addition of new species by extrapolation in relation to sampling effort) also were applied (Palmer 1990, Soberón & Llorente 1993), and included Clench's (Clench 1979) and Linear Dependence (von Bertalanffy 1938) models, as well as the slope of the cumulative species curve to assess the completeness of assemblages (Jiménez-Valverde & Hortal 2003, Hortal & Lobo 2005).Slopes were obtained by means of the first derivative of Clench's and Linear Dependence functions (Novelo-Gutiérrez & Gómez-Anaya 2009).

Classification and ordination:
Cluster Analysis (CA) on a Bray-Curtis (BC) similarity matrix with unpaired group mean average (UPGMA) amalgamation was used to explore faunal similarities among streams and seasons.This analysis was performed using PC-ORD v4.5 (McCune & Grace 2002).Additionally, a Canonical Correspondence Analysis (CCA) (CANOCO for Windows v4.5) was used to relate transformed species abundance [log(y+1)] to environmental variables (ter Braak 1986).The number of environmental variables was then reduced using the automatic forward selection option in CANOCO.Variables explaining most of the variation in the data were used to construct the final model.The statistical significance of the relationship between species and environmental variables was tested using a Monte Carlo permutation test (n=499), with an F-ratio as the sum of all eigenvalues for the test statistic (ter Braak & Prentice 1988, ter Braak & Smilauer 2002).
There was a significant difference among streams (one-way MANOVA: Wilk's lamb-da=0.46,F 16, 141 =2.53, p<0.05) for width, depth, current velocity and discharge.The Bonferroni paired t-tests showed differences in depth and discharge for TZ with CH (Chichihua), EZ (Estanzuela) and RP.There was a significant difference in width between TZ and CL, while significant differences in current velocity were observed for TZ with EZ and RP.
A two-way MANOVA showed a significant effect of stream (F 16, 480.3 =228.8,p<0.05), seasons (F 12, 415.7 =182.2, p<0.05) and their interaction (F 48, 606.8 =17.6, p<0.05) on physicochemical variables.A Bonferroni paired t-test showed significant differences (p<0.05) in temperature among streams, except that of RP to EZ and TZ.Sites CL and CH exhibited the lowest average temperatures, 20.14ºC and 22.62ºC, respectively, while the highest were those of TZ (29.35 ºC) and RP (28.03ºC).Most of the Bonferroni contrasts were significant for pH, except those between CH-EZ and CL-TZ, and all sampling sites had a basic pH.The Bonferroni contrasts for temperature among seasons showed significant differences between autumn-spring, autumn-summer and springsummer.The average winter temperature was lower than that for the other seasons (Table 2).Most of the paired comparisons for conductivity showed significant differences among sites (p<0.05),except those between EZ and CH, and between RP and TZ.CL exhibited the lowest average value of all streams (50.08µ/cm).There was no significant difference in conductivity among seasons (p>0.05).The Bonferroni contrasts for oxygen showed significant differences between TZ and the other four streams, and between EZ and CL, while there was no difference between RP-CL, RP-EZ, or between CH-CL and EZ-RP.The average oxygen concentration of TZ was lowest (4.31ppm) compared to other sites.The contrasts indicated that the seasons are very different for oxygen concentration, except during spring-summer.
Diversity: H' for CM was 4.11, higher than any stream individually, and ranged from 1.85 (CL) to 3.31 (CH).CH and EZ had the highest site values (3.31 and 3.13, respectively), J ranged from 0.43 (TZ) to 0.67 (CH and EZ), D from 0.16 (EZ and CH) to 0.47 (CL), and R was highest in TZ (4.44), and lowest in CL (2.57) and EZ (2.84).According to Renyi's profiles, diversity at CH was the highest and that for CL the lowest.The order of diversity by stream was CH>EZ>RP>TZ>CL (Fig. 2).
Renyi's diversity profiles showed that diversity declined from spring>summer>autumn>winter.

Non-parametric estimators of gamma richness:
The estimates using Chao2, Jack2, Bootstrap and MaoTau were 103, 109.8,84.5, and 76 species, respectively, yielding sampling efficiencies of 72.8%, 68.3%, 88.8% and 98.6%, respectively (Table 4).These estimators predicted 9 to 34 species were missing from the list for the CM.The upper MaoTau limit predicted 85.4 species (87.8% efficiency) with 10 potential additional species.However, the non-parametric richness estimators that used abundance data (Chao1, ACE) estimated 91.2 and 94.1 species, respectively, yielding Parametric models: The Clench function fits the data better for each stream and for the CM (Fig. 3, Table 5), with R 2 ranging from 95.13% (RP) to 99.71% (CH) among streams, and being 98.37% for the CM.The theoretical number of missing species ranged from 2 (EZ) to 13 (TZ) among streams, and 8 for the CM.Curve slopes ranged from 0.04 (EZ) to 0.26 (TZ) among streams, and 0.04 for the CM.
Cluster analysis: Colorín (CL) separated first at a major distance (Fig. 4), followed by Ticuiz (TZ).The CL fauna was the poorest,  where Hetaerina capitalis dominated (almost 75% of the total abundance), and with many (55.56%)rare species (species with relative abundance less than 1%).TZ was the stream with the largest number of observed species, most (80.56%) of which were rare.Overall turnover rate was less than 25%, except in EZ-RP (48%).Considering the seasonal assemblages, the cluster analysis shows the formation of two groups of contiguous seasons: the first one formed by spring+summer and the second one for autumn+winter.The first group had the assemblages with the highest number of species and abundance, where the dominant species were the zygopterans Argia pulla, H. capitalis and A. oenea; while in the second group the anisopterans Macrothemis pseudimitans and Erpetogomphus elaps were dominant.

Species-environment relationships:
The results of CCA were globally significant (trace=2.66,F=4.04, p=0.002,Table 6).The first three axes accounted for most of the variability (75.3%) for physiochemical variables and species abundance (inertia=19.60).The first canonical axis was statistically significant (eigenvalue=0.945,F=11.90, p=0.002) and explained 35.5% of the variance, while the second and third axes explained 22.4% and 17.4%, respectively.The most important environmental variables explaining larval variation were determined from axis correlations, with gradient and elevation exerting the strongest influence on axes 1 and 2, and axis 1 also influenced by conductivity to yield three putative groups (Fig. 5).

DISCUSSION
As commonly occurs in Mexico, more Anisoptera (62.67%) were observed than Zygoptera (37.33%), and the Libellulidae and Coenagrionidae predominated (19 and 29 species, respectively) with the genus Argia having the most species (14) (Alonso-Eguía Lis 2004, Gómez-Anaya et al. 2000, Novelo-Gutiérrez & González-Soriano 1991, Novelo-Gutiérrez et al. 2002, Bond et al. 2006).However, some variation in the proportion of Anisoptera to Zygoptera species was observed: while the number of observed Anisoptera species was similar to that for Zygoptera in CH, CL and EZ, in RP and TZ five and nine more Anisoptera species were recorded, respectively.
The extreme differences in number of species presented by TZ and CL can be explained because TZ possesses a mixed lotic-lentic environment, is low in altitude, open and sunny, has floating water lilies, Eichhornia crassipes (Mart.Solms), and submerged (Egeria sp.) macrophytes, and is exposed to continuous human impact.Our results could be explained by the intermediate disturbance hypothesis (Connell 1978) which states that diversity is low after a disturbance, when only a few species have survived or few colonizing species dominate under the new environmental conditions.Diversity is high when disturbances occur at an intermediate frequency or with intermediate intensity.In this respect, frequent dragging in the TZ channel to extract gravel can be regarded as a disturbance since it alters the habitat.Furthermore, dragging generates medium sized lentic habitats which are mainly occupied by opportunistic libellulids such as Erythrodiplax, Erythemis, Miathyria, Micrathyria, Orthemis, and Perithemis, increasing TZ diversity.
CL, on the other hand, is located in a ravine at medium altitude, is small in size and almost fully covered by riparian vegetation, is a stream structured as successive terraces, and has negligible human impact.These results support other studies showing species richness to be positively correlated with stream size and canopy openness (Kinvig & Samways 2000, Clausnitzer 2003, Dijkstra & Lempert 2003, Campbell et al. 2010).
The number of species among streams, and their abundance and numeric dominance also exhibited temporal variation.In the dendrogram generated by the CA, abundance and number of species were highest in the spring+summer assemblage, and were dominated by the zygopterans Argia pulla, H. capitalis and A. oenea, while the autumn+winter group had the lowest number of species and abundance was dominated by the anisopterans Macrothemis pseudimitans and Erpetogomphus elaps.The alteration in larval dominance may be related to the presence or absence of predators, with dominance by large anisopterans when predatory fish are absent, and by smaller zygopterans when predatory fish are present (Mikolajewski & Johansson 2004).Large anisopteran larvae are more conspicuous to predators than smaller zygopterans.
Our ordering of diversity was based on Renyi's profiles (Tóthmérész 1995) which have been proposed as an efficient method for comparisons (Southwood & Henderson 2000), although infrequently used as such on the diversity of aquatic macroinvertebrate assemblages (e.g.Sipkay et al. 2007), and even less so for Odonata assemblages (e.g.Jakab et al. 2002).This lack of use may occur because this method involves more calculations than that for simple diversity indices (Tóthmérész 1995).The idea of ordering diversity was first proposed by Patil & Taillie (1977, 1979) and Solomon (1979).By adding a new dimension to the ecological approach of diversity, it connects with the implications of proposals for diversity conservation that focus on recognizing areas of high diversity.We found this tool useful in comparing the diversity of Odonata larval assemblages within the CM.Using this method, we have shown that diversity at CL was the lowest, and that for CH the highest, with the latter being an ecologically preserved area in the state of Michoacán.Among habitats, Renyi's profiles showed that the highest diversity occurred along stream margins, compared to riffles or pools, even when stream margins and riffles were almost equal in abundance, 4 044 and 3 905, respectively.Additionally, different dominant species were observed among habitats, with E. elaps mainly recorded from riffles and M. pseudimitans and A. pulla mainly from stream margins.Finally, Renyi's profiles showed a gradual decrease in seasonal diversity, spring>summer>autumn>winter, without intercrossing of profiles.
Beta diversity has been poorly evaluated for Mexican invertebrates (Soberón et al. 2005, Escobar 2005, Favila 2005, Noguez et al. 2005).Species turnover is the least studied and understood component of diversity, even though conservation efforts promote great interest in its evaluation (Gaston & Blackburn 2000).There is an inverse relationship between regional beta diversity and the species distributions in a region (Harrison et al. 1992).If the species from a given region have small areas of distribution, such that sites differ in species composition, then a high beta diversity results.However, if species are widely distributed over most of the region, then sites will be more similar to each other in species composition, resulting in low beta diversity (Arita & León-Paniagua 1993, Scott et al. 1999; see also Campbell et al. 2010).In our study, beta diversity was evaluated as the complement of BC, a key element in understanding the relationship between the diversity in the CM (regional or gamma diversity), and diversity of each stream at the local, or alpha level (Cornell & Lawton 1992, Ricklefs & Schluter 1993).Most species showed a restricted distribution in the CM, which suggests a high turnover rate, a conclusion supported by the fact that most of the Bray-Curtis paired similarities were smaller than 25%.Hence, beta diversity is an important component of gamma diversity among these sites (Novelo-Gutiérrez & Gómez-Anaya 2009).Our results also support the betadiverse Mexico hypothesis (Arita & León-Paniagua 1993, Sarukhán et al. 1996), a result of high environmental heterogeneity in the CM.From the point of view of biological conservation, beta diversity is an important component that should be taken into account in the establishment of efficient conservation strategies of particular natural areas and/or species (Scott et al. 1999).
Considering that observed gamma diversity for the CM comes from sampling, we estimated the theoretical true gamma by using several non-parametric and parametric richness estimators.We used Clench's model because it has been used often and it better fits this kind of data.For the non-parametric estimators we used both incidence and abundance data, because there is no strong recommendation for a particular one.According to the sampling efficiency Chao1 performed better for EZ, CH and RP, while Bootstrap performed better for TZ, CL, and over all sampled sites.On the other hand, Clench's model better fit the data based on the coefficient of determination (R 2 ), and we assessed the completeness of our Odonata larval lists using the slope of the curve at maximum effort for the extrapolation estimators provided by the first derivative of both the Clench and Linear Dependence functions.According to these values, the most complete lists were those with small slopes, where the increasing rate of species additions tended toward zero.These values agreed well with the fact that sampling effort was highest at EZ (n=65) where the lowest slope was found to be associated with the highest sampling efficiency and just two missing species.On the other hand, the number of species collected at TZ yielded 73.22% efficiency and 12 potentially missing species.These observations agree with the fact that a negative and significant correlation between the number of species collected and slope was observed (r=-0.76,p<0.05), and also between slope and percentage efficiency (r=-0.90,p<0.05).
Our CCA results show that larval abundance is primarily associated with elevation and stream slope.This observation is consistent with Hofmann & Mason (2005) and Sato & Riddiford (2008) who observed that current velocity is one of the most important factors related to the distribution and abundance of Odonata.As gradient (slope) increases with elevation, so does current velocity, leading to reduced species richness.Several studies have revealed odonate species richness declines with increasing altitude (Borisov 1987, Samways 1989, Vick 1989, Campbell et al. 2010), and in some places few or no Odonata occur above 2 000m (Laidlaw 1934).However, the size of a regional assemblage over a broad spatial area is a function of slope, latitude and altitude (Corbet 1999), as well as local factors such as stream morphology, aquatic vegetation, riparian coverage and physicochemical variables.The separation of CL first in Fig. 5 could be a result of such local differences.Among the sites surveyed, we found most species at 10m.a.s.l.(TZ), and the fewest at 1 050m.a.s.l.(CL).Campbell et al. (2010) calculated a Pearson correlation coefficient between elevation and species richness for all CM windward sites as r=-0.92(p>0.05).Even though they revealed a decline with increasing elevation, the marginal insignificance of the correlation was due to the low number of sites surveyed (n=3).Yet, despite the marginal insignificance, average taxonomic distinctness (a measure of phylogenetic diversity; e.g.Warwick & Clarke 1995, Clarke & Warwick 1998) was significantly and negatively correlated with observed species richness, suggesting the decline in species richness with elevation is real.It is known that altitude-related changes in species composition usually entail a progressive disappearance of the more abundant species at higher altitudes, although some taxa can have their centers of distribution at high and intermediate altitudes (Corbet 1999).In the CM we observed Hetaerina capitalis as the most common species at CL (1 050m, see Table 2), decreasing notably in larval abundance at CH (1 130m), while other species also present at CL such as Progomphus zonatus, Brechmorhoga tepeaca and Macrothemis ultima also decreased or disappeared in CH.Larvae from these four species were not found at lower elevations.We also observed that a higher number of species of Gomphidae (10 out of a total of 13), as well as most of the abundance of this family (more than 90%, but primarily from Erpetogomphus elaps) was recorded at intermediate altitudes, between 408m.a.s.l. and 560m.a.s.l.Notwithstanding, Erpetogomphus elaps was the most abundant and widespread species in the CM, its highest larval abundance occurred at 152m.a.s.l.On the other hand, Libellulidae were better represented with 22 species (from a total of 29) below 560m.a.s.l.Only 12 species from this family were recorded as larvae above 1 000m.a.s.l.; Brechmorhoga tepeaca, Dythemis multipunctata, Macrothemis ultima and Paltothemis cyanosoma were recorded exclusively above 1 000m.a.s.l.In contrast with Hoffmann (1991) who reported a predominance of species of Rhionaeschna (Aeshnidae) at higher elevations in the Peruvian Andes, this family was not well represented in the CM either in number of species or abundance.While most Coenagrionidae species were recorded at CH (11 out of a total of 19); two species, Argia cuprea and A. lacrimans, were exclusively recorded from 1 050m.a.s.l.(CL).
Finally, recommendations for the conservation of odonate diversity in the CM are based on the high beta diversity observed and on stream quality; as shown in the CCA analyses where it was possible to incorporate characteristics of local microhabitats.The conserved vegetation cover and the negligible human impact in CL are, for example, important factors to consider (Buchwald 1992, Remsburg et al. 2008).Alteration of plant cover could change the aquatic microclimate and lead to changes in Odonata diversity (e.g.Samways & Sharratt 2009, Campbell et al. 2010).The CL fauna, although minor in assemblage size, comprises a particular group of infrequent species: Argia cuprea, A. lacrimans, Aeshna williamsoniana and Paltothemis cyanosoma, all of which occurred uniquely in this locality.Other species such as Hetaerina capitalis, Erpetogomphus cophias, Progomphus zonatus, Brechmorhoga tepeaca and Macrothemis ultima have their maximum and almost exclusive records at this site.It is feasible that their association with stream slope may be based on associations with conserved canopy, small size, shallow runs, small to moderate sized pools and sand banks with rough grains.This opens some questions: What is the importance of the ravines like CL as shelters for these infrequent species in a broader scale (e.g. in all the Michoacán Mountain Ranges?) Are the odonate assemblages different in each ravine increasing the beta diversity?Our results support the suggestion for locally based conservation efforts along the windward transect (e.g.Campbell et al. 2010), and support proposals describing the CM as a biologically rich area for conservation.

ACKNOWLEDGMENTS
Financial support was provided to RNG by Comisión Nacional de Ciencia y Tecnología (CONACyT) grant 43091-Q.

Fig. 1 .Fig. 2 .
Fig. 1.Total numerical dominance of Odonata larvae collected from the five sampling sites in the Coalcomán Mountain Range.Only the 48 most abundant species have been plotted (species key in Table2).

Fig. 3 .
Fig. 3. Cumulative species curves for Coalcomán Odonata larval assemblage generated by the Clench and Linear Dependence functions.The dashed line is the asymptote estimated by the Clench model.

Fig. 5 .
Fig. 5. CCA biplot showing environmental variables (arrows) most strongly correlated with axes CC1 and CC2.Variables include elevation, gradient, conductivity, season, pH, oxygen, temperature, and habitat.Environmental variables providing strong influence on larval assemblage composition have longer arrows than less influential variables.

TABLE 1
Annual means and ranges (Min -Max) for the stream physicochemical variables

TABLE 5 Sites
, number of species, parameters and predictions of two species accumulation models fitted for each odonate larval assemblage (key to sites in Table1) obs =number of observed species, a=the slope to the beginning of sampling, b=a parameter related to the mode of accumulation of new species during sampling, a/b=asymptote, n=total number of samples, R 2 =coefficient of determination, eff=percentage of efficiency, and sl=slope.The slope was evaluated using the first derivative of Clench's function [a/ (1+b*n) 2 ] and the first derivative of the Linear Dependence function [a * exp(-b*n)].

TABLE 6
Results of CCA of log-transformed larval abundance as a function of their environmental variables Cluster analysis dendrogram based on Bray-Curtis' similarity (using the UPGMA Linkage Method) of Odonata larvae collected from the Coalcomán Mountain Range.EZ=Estanzuela, TZ=Ticuiz, RP=Pinolapa, CH=Chichihua and CL=Colorín.