Rev. Biol. Trop., 44(2): 581-590,1996

Vegetation-environment relationships based on a life-forms c1assification in a semiarid region of Tropical

Zavala-Hurtado, J.A.!, P.L. Valverde!, A. Díaz-Solís2, F. Vitel and E. Portillal 1 Departamento de Biología, Universidad Autónoma Metropolitana- Iztapalapa. Ap. Postal 55-535, México 09340, D.F. México, e-mail [email protected] Jardín Botánico y Vivero de Cactaceas "Dra. Helia Bravo-Hollis",Zapotitlán, , México.

(Rec. 15-V-1995. Rev. 23-VI-1995. Acep. 28-VI-1995)

Abstract: Taking the view that morphological characteristics represent behavioral strategies of to cope with environmental pressures, we defined 18 life-forrns, using multivariate classification techniques in a tropical semiarid ecosystem in Central Mexico. A multiple discriminant analysis confirrned the existence of these groups. A nul! model of random rnembership of to life-forrns was significantly different from our classification. Vegetation-envíron­ ment relationships were exanúned wíth Detrended Canonical Correspondence Analysis (DCCA). Ordínation axes were ínterpreted mainly by altitude and soíl moisture. Response curves of life-forrns along these gradients were explored fit­ ting generalized linear models (OLIM's). We believe that the life-forrns approach for the study of vegetation-envíron­ ment relationships ís a valid alternative to the tradítional specíes approach usually used in phytosociological research because: í) lífe-forrns number was found to be an excellent specíes diversity predictor, ii) this approach enables consid­ erable reduction in the bulk of data without losing ecological inforrnation, and iíi) life-forrns represent ecological strategies per se and, they constitute an index of the number of different ways the desert's resources are utilízed.

Key words: Morphological attributes; morphologica1 strategies; ordination; multiple discriminant analysis; response curves; xerophyllous scrub; desert plants.

The occurrence of structural similarities ber and nature of the morphological traits used among desert plants, independent of their phy­ and in the number of life-form groups pro­ logenetic relationships and geographical distri­ duced. bution, is well known . There is a preponder­ In this paper we describe a multivariate clas­ ance of certain growth forms in desert floras sification of 107 perennial species based that are rare or restricted in other types of on morphological characteristics in a semiarid ecosystems. Nevertheless, the causes of these region of tropical Mexico. The basic assump­ presumed convergences are not well únder­ tion is that plant species would be naturally stood (Solbrig et al. 1977, Bowers and Lowe clumped in a few morphological categories 1986, Cody 1989). A fundamental problem is each of which represents a convergence of to understand how the environment acts as a strategies to cope with the desert environment. selection pressure on the shape and function of Our aim was to describe vegetation-environ­ plants. How do desert plants coexist interacting ment relationships based on these life-form cat­ with each other and with the physical environ­ egories. ment? Vegetation-environment relationships are There have been several attempts to define usually studied along gradients. The concept of morphological strategies of desert plants in environmental gradients has been a comerstone relation to environmental pressures (Shreve in the development of ecological theory 1942, Cody 1989, Leishman and Westoby (Oldand 1992). Austin et al. (1984) recognize 1992). These attempts vary widely in the num- three basic environmental gradients types: (i) 582 REVISTA DE BIOLOGIA TROPICAL indirect gradients, those which do not have a The study sÍte utilized for the vegetation­ direct influence on plant growth (e.g., altitude); envíronment relationships is a protected area (ii) direct gradients, those which have a direct adjacent to the '' Botanic influence on plant growth, but are not resources Garden, located in the middle of the Zapotitlan potentíally subject to competitíon (e.g., pH); Valley, 28.5 Km SW of Tehuacan city. The and (iii) resource gradients, those environmen­ study site occupies an area of 1 Km2 and con­ tal variables which are direct resources influ­ sists of the tetechera vegetation unit. encing plant growth (e.g., soil nutrients). Using Five 10 x 10 m permanent plots have been certain assumptions, direct environmental gra­ established at each site where vegetatíon dients can be regarded as dimensions of the dynamics studies are carried out. In these 15 Hutchinsonian niche concept. Following this, plots we recorded 107 perenníal plants. We species response curves along direct gradients selected ten individuals of each species in wouId represent their realized niches in one every 10 x 10 m plot that it was present, and dimension (Austin et al. 1990) and, hence, recorded the presence or absence of 30 mor­ would provide measurements of niche ampli­ phological attributes (Table 1). These attributes tude and overIap. have clear adaptive significance in terms of Taxonomical nomenc1ature in this paper fol­ photosynthesis optimization (presence of lows Davila et al. (1993). leaves, photosynthetic stems, etc.), water stor­ age and conservatíon (waxy leaves and stems, succulence, etc.), and thermic regulation (mor­ MATERIAL AND METHODS phology of stems, hairy leaves, etc.).

Our study site was located in the semiarid TABLE 1 valley of Zapotitlán (18° 20' N, 9T 28' W), a local basin in the Pueblan- Oaxacan region in Morphological attributes usedjór classification and multiple discriminant analysís the Mexican State of Puebla (Vite et al. 1990). This is a unique regíon because of its biological richness. About 30% of its species are endemic 1 Spines present 16 Branched and it is especially rich in columnar cacti 2 Succulent stem 17 Extensive branching (Villaseñor et al. 1990). 3 Exfoliant cortex 18 Plagiotrophic branching The c1imate is dry with summer rains. 4 Perennial stem 19 Wide base Annual mean temperature is 18-22°C and pre­ 5 Spiny stem (*) 20 Leaves present (*) cipitation is around 400 mm/yr. The soils are 6 Photosynthetic stem 21 Rosetophyllous leaves (*) shallow, stony, and halomorphic (Byers 1967). 7 Waxy stem 22 Simple leaves Arid conditions are produced by the rain shadow 8 Woody stem 23 Microphyllous leaves of the Sierra Madre Oriental. The vegetation has 9 Ereet stem 24 Hairy leaves been c1assified as xerophyIlous scrub 10 Flattened stem 25 Succulent Ieaves (Rzedowski 1978) or as thorn scrub I1 Solitary stem 26 Glabrous leaves (Smith 1965), and is a well-preserved example 12 Globose stem 27 Waxy leaves of this vegetation type, that supposedly covered 13 Candelabrous shape (*) 28 Perennialleaves the region sorne 10,000 years ago (Smith 1967, 14 Ribs present 29 Caudex present Zavala-Hurtado 1982). Zavala-Hurtado (1982) 15 Tubercles present 30 Epiphytic habit described four vegetation units in the Zapotitlan Valley: thorn scrub (dominated by thorny shrubs (*) Attributes not entered by multiple discriminant analysis. and trees, mainly legumes, agavaceae and low cacti), cardonal (thorn scrub with the columnar On the species-attributes matrix (107 x 30) cactus Cephalocereus columna-trajani Weber), we carried out an agglomerative cluster analy­ izotal (thorn scrub with Yucca periculosa Baker sis usíng Ward's method (1963) with the or Beaucarnea gracilis Lem., and tetechera SPSS+Pc package (Norusis 1988). The number (thorn scrub with the columnar cactus of groups in the species c1assification was cho­ Neobuxbaumia spp.). There are permanent sites sen subjectively, based on a visual inspectíon for research on vegetatíon dynamics in the of the dendrogram obtained (Fig. 1). Each of tetechera, izotal and cardonal units. these groups was regarded as a life-form (LF). ZAVALA-HURTADO et al.: Vegetation-environment relationships 583

Rescal.ed Di.stance. Cl.uster CoJablne LIFE-FORK SPECIES o 5 10 1.5 20 [ I I I I 79 101 24 LO' 1. 77 76 90 [ .49 80 81 83 LF 2 106 84 4 82 39 44 35 LO'3 2S 78 f 100 85 65 67 LO'4 60 62 66 43 45 LF5 63 64 61 LF6 f 1� [ 12 l 38 1i LF 7 10 8 48 102 LF 8 [ 103 98 52 LF 9 71 70 S7 96 55 56 LF10 32 3. 28 30 3>- 29 .. 58 40 LF11 69 5 92 27 36 10. LF 12 105 50 [ 94 95 23 88 5' 3 75 LF 13 93 22 99 68 42 2 74 73 91 1 26 19 37 20 34 54 21 72 86 LE' 16 87 41 16 17 LF17 14 18 15 59 97 53 LF18 47 89 �]- 46

Fig. 1. Dendrogram with the agglomerative schedule oC the c1assificationof 107 species from the semíarid valley of Zapotitlan, Mexico into 18life-forms (indicated by LF#) using Ward'smethod. See App. 1 for speciesnames. 584 REVISTA DE BIOLOGIA TROPICAL

Results of the classification were evaluated Soil moisture was estimated as a percentage by a stepwise multiple discriminant analysis of the soil dry weight (A very and Bascomb using SPSS+pc. This method was also applied 1974). pH was measured in a water- saturation to a null model of random assignment of percentage preparation (Jackson 1958) using a species to the same number of groups detected pH-meter Chandos type M43. with the cluster analysis. The objective of this Altitude, aspect and slope were measured comparison was to test the nulI hypothesis of using conventional methods. Topography (con­ random membership to life-forms. cave, plain and convex) and stoniness were The null model was constructed permuting estimated visualIy. at random the life- form constituency of each With these data, we built-up two matri­ species (maintaining its own morphological ces: life-forms x samples and environmental attributes). variables x samples. Both matrices were ana­ Analysis of vegetation-environment rela­ lyzed simultaneously with a Detrended tionships was carried out in the protected area Canonical Correspondence Analysis (DCCA) of the 'Helia Bravo Hollis' Botanic Garden using the CANOCO package (ter Braak 1987). where we set-out a grid of one hundred 100 x To explore the relationship between the 100 m squares in an area of 1 Km2. In the defined life-forms and the environmental vari­ southeast comer of each of the 100 resulting ables considered, we made a biplot of the life­ squares we located a 5 x 5 m plot. As 22 edge formsordination on the sample space. The vec­ sites were occupied by cultivars, we were left tors show direction and relative importance of with 78 5 x 5 m plots. Within each plot we the environmental variables significantly corre­ recorded presence/absence data for perennial lated with the sample ordination axes. plants, and took into account their life-forms The response of different life-forms to envi­ membership according to the c1assification ronmental gradients was estimated by fitting analysis. AIso, we recorded 11 continuous and Generalized Linear Models (GLIMs, Baker and 12 categorical environmental variables for each Nelder 1978) to the defined life-forms. plot (Table 2). Independent variables (gradients) tried were the Incident radiation was estimated using the environmental variables that explained signifi­ computer package INSOL (Sánchez-CoI6n cantly the sample ordínation axes. This 1987 unpubl.) considering geoposition of the approach is usefui to predict the probability of plots and topographical interferences. a life-form being present (in a quadrat of 5 x 5 m, in this case) at a given point along the gradi­ ent. We assumed a binomial distribution for TABLE2 presence/absence data and used a logit transfor­ Environmental variables used for ordinationby DCCA mation to link this distribution with the linear predictor (Austin et al.1984). Continuous Categorical

1 Altitude m.a.s.'. 1 North eastem aspect RESULTS 2 Winter incident 2 South-eastem aspeCI radiationMJ/m' Cluster analysis grouped the 107 species to 3 Springincident 3 South-westem aspect 18 life-forms (Fig. 1). AH morphological attrib­ radiationMJ/m' 4 Summer incident 4 North-westem aspect utes listed in Table 1 were used for the defini­ radiation MJ/m' tion of life-forms. The dendrogram in Fig. 1 5 Fall incident 5 Indeterminate aspect shows the agglomeration schedule. Two large radiationMJ/m' groups are evident: species with (life-forms 1 6 Yearlyincident 6 Concave topography to 5) and without leaves (life-forms 6 to 18). radiation MJ/m' 7 Soil moisture at 7 Plain topography AH the latter life-formshave succulent stems. 10 cm depth (%) The first group of life-forms consists entire­ 8Soil moisture al 8Convex topography ly of members of the Cactaceae family (31 20 cm depth species). LFl (seven species) inc1udes 9 Soil pH at 10 cm depth 9 Stoniness (0-25%) 10 Soil pH at 20 cm depth 10 Stoniness (25-50%) branched columnar and cande1abrous-like 11 Slope (") 11 Stoniness (50-75 % ) species, LF2 ineludes seven species with 12 Stoniness (75-100%) branched flattened stems, columnar non- ZAVALA-HURTADO et al.: Vegetation-environment relationships 585 branched and barrel-like cacti are clumped in di&criminamIullClion l! LF3 (seven species), whereas five species of solitary globose cacti (tubercles present) belong to LF4, and LF5 is formed by five species of gregarious globose plants (tubercles present). A single case of misclassification was found in this group: Coryphanta palida Brit. & Rose a solitary globose cactus was located in LF3 bearing more morphological affinity with members of LF5.

The secónd outstanding group (leaves pre­ -20 -10 3D sent), with 77 species, is subdivided in two discriminamfunction 1 further groups: species with succulent Ieaves Fil. 2. Centroids andpoligonal territories of 18 life-forms (LF's 6 to 8) and with non-succulent leaves in discriminan! space given by the fírst and second func­ tions of multiple discriminant analysis. (LF's 9 to 18). Within the life-forms possess­ ing succulent leaves there is LF6 (four species) axis 2 formed by arboreal rosette-Iike plants, LF7 300 -----, - . ,�----��-----, (six species) with acaulescent rosette-like I species, and LF8 (three species) with small 200� ¡ e, rosette-Iike plants (sorne of them epiphytes). The group with non-succulent leaves is com­ 100 1- \'" posed of a variety of trees and shrubs with and \r ,. s. j o:I " without spines. Life form 9 (three species) �ALT "" consists of climbing species, LFIO (ten I -100 i species) of short shrubs with unarmed simple ! '" I leaves, LFll (six species) of medium-sized l''-_--L_---' __ _ __'__ -200 ��� __._1 shrubs with microphyIlous leaves. The LF12 -200 -100 O 100 200 300 400 500 axis 1 (seven species) incJudes short shrubs with Fig. 3. Biplot of life-tonns orOmatlon in the sample space hairy leaves. Unarmed trees and shrubs, main­ with respect to four enviromental variables (Iines). Iy with composed leaves were grouped in CT=concave topography; ALT=altitude; SEA=south-east­ LF13 (ten species), whereas LF14 (seven ernaspect; SM=soil moisture at 20 cm depth. species) incJudes thorny trees and shrubs. Medium sized unarmed shrubs with simple discriminant function separates these two leaves are in LFI5. Species with ephemeral groups by the presence or absence of succulent leaves, succulent photosynthetic and unarmed stems, spines, ribs and tubercles. stem were grouped in LF16 (three species). The Multiple Discriminant Analysis of the LF17 (five species) includes trees and shrubs null cJassification model, revealed that only wi th an exfoliant cortex, and LF 18 (six 15.89% of the species were classified correctIy. species) consists of shrubs with simple waxy Our life-forms classification differed signifi­ leaves. cantly from this random assignation (Chi Multiple Discriminant Analysis validated square = 1414.26; P < 0.001). this 18 group cJassification (Fig. 2), with Fig. 3 shows the ordination of life-forms on 99.07% of the species being correctly classi­ axes 1 and 2 of DCCA. These axes accounted fied. The first two discriminant functions for 60.54% of variance (37.77% and 22.99%, accounted for 55.25% of variance (39.99% and respectively). The same Figure shows the biplot 15.26% for discriminant function 1 and dis­ of life-forms ordination on the sample space and criminant function 2, respectively; p < 0.001). the direction and relative importance of the most The structuring of groups shown in the dendro­ influential environmental variates. Axis 1 was gram of Fig. 1 is clearly depicted in the dis­ significantly explainedby altitude (r2= 0.55; � < criminant space of Fig. 2. Although, according 0.001) and soil moisture at 20 cm depth (r = to the multiple discriminant analysis, segrega­ 0.44; p< 0.001). Axis 2 was explained by con­ 2 tion between the two maín groups is not given cave topography (r = 0.80; p < 0.001) and by the presence or absence of leaves. The first south-eastern aspect (r2= 0.19; p < 0.001). On 586 REVISTADEBIOLOGIA TROPICAL

It can be seen that climbing deciduous plants axis 1 200� (LF9), unarmed shrubs with succulent stems I and ephemeral leaves (LF16), trees and shrubs with simple leafs and waxy stems (LF18), 150 f acaulescent rosette-like succulents (LF7) and ! trees with microphyllous leaves and exfoliant I 100 i stems (LF17) would be restricted to relatively high altitude and/or relatively high water soil I content. On the other hand, unarroed trees with 50 l· Y'39.BS46' •. 3'032(x) 1 "'0,"" I composite leaves (LF13) and small globose gre­ I •• 1'.0.001 I garious spiny succulents (LF5) would be found O ���, ----�' _._----.L ����...... J 0246 8 ro � in lower and less humid sites. Succulents with number 01 life-forms spiny flattened photosynthetic stems (LF2), Fig. 4. Relationship between ordination axis 1 (exp. unbranched columnar cacti (LF3) and small var=37.77%)and life-forms diversity. rosette-like epiphytes (LF8) occupy moderately xeric sites. The other eight life-forms are a expected to be found in intermediate conditions or to have a wider range of tolerance. Fig. 5 shows significant (p < 0.05) response curves of eight representative life-forms to (a) altitude and (b) soil moisture at 20 cm depth, the most important environmental variables in the determination of sample ordination. Climbing deciduous plants (LF9) show low probabilities of occurrence along both gradients and are restrícted to moderately high soil mois­ ture contents and relatively high altitudes. 1350 1370 1390 1410 1430 1450 1470 1490 Unarmed shrubs with succulent stems and altltude ephemeral leaves (LF16) are restricted to high altitudes. Rosette-like acaulescent succulents b (LF7), unarmed trees with composite leaves (LF13) and small rosette-like plant (sorne of them epiphytes, LF8) raise their probabílity of occurrence as both altitude and soil moisture increase, but become rarer in the upper part of the altitude gradient. Succulents with spiny fIat photosynthetic stems (LF2) show bimodal response curves for both gradients with rela­ tively high probabilities of occurrence at both ends of the soil moisture gradient and at low altitudes. Finally, short deciduous unarmed shrubs (LFlO) and small globose gregarious o 8 m W M m 18 20 spiny succulents (LF5) show wide bell-shaped Boil moisture at 20 cm depth response curves with highest probabilities of occurrence in the middle of both gradients. Fig. 5. Significant (p

1985, 1987, Austin et al. 1984, Austin et al. un ecosistema semiárido del trópico mexicano. Un análisis 1994), but the interpretatíon of the ordination discriminante múltiple confirmó la existencia de estos 18 axes did allow us to get an insight into the per­ grupos. Esta clasificación fue significativamente diferente de la esperada por una asignación aleatoria de formas de formance of life-forms along explicit envrron­ vida. Se exploraron relaciones vegetación-medio ambiente mental gradients using a direct gradient analy­ utilizando un análisis de correspondencia canónico sin ten­ sis approach. dencia (DCCA). Los ejes de ordenación fueron interpreta­ The simple quadratic models used revealed dos principalmenmte en términos de altitud y humedad del suelo. Se exploraron las curvas de respuesta de las formas significant bell- shaped (mainly platykurtic, and de vida a lo largo de estos gradientes ajustando modelos li­ bimodal in sorne cases) responses of life-forms neales generalizados (GLIM). Consideramos que el plan­ to the environmental variates examined. Most teamiento de formas de vida para el estudio de las relacio­ of the life-forms response curves were wide, nes vegetación-medio ambiente es una alternativa válida al tradicional uso de categorías taxonómicas en estudios fito­ perhaps because theyencompass realized niches sociológicos porque: (i) se encontró que el numero de for­ of their constituent species. In this sense, life­ mas de vida es un buen predictor de la diversidad de espe­ form response curves would be a guild level cies, (ií) este planteamiento permite una reducción conside­ analogous to the qualitative environmental real­ rable de la base de datos sin pérdida de información ecoló­ gica y (iii) las formas de vida representan estrategias ecoló­ ized niche (QERN) of species (Austin et al. gicas per se y constituyen un índice del número de formas 1994, Austin et al. 1990). Nevertheless, life­ diferentes en que son utilizados los recursos del desierto. forms response curves should not be regarded simply as cumulative species response curves REFERENCES because these categories may behave in a quite different way to the same environmental gradi­ Austin, M.p. 1980. Searching for a model fOf use in vegeta­ ent (Okland 1992). tion analysis. Vegetatio 42: 11-21. The lífe-forms approach for the study of veg­ Austin, M.P. 1985. Continuum concept, ordination meth­ etatíon- environment relationships is considered ods and niche theory. Ann. Rev. Ecol. Syst. 16: 39-61. a valid alternative to the traditional species approach usually used in phytosociological Austin, M.P. 1987. Models for the analysis of species' research: firstly because, life-forms number was response to environmental gradient:s. Vegetatio 69: 35-45. found to be an excellent predictor of species Austin, M.P., R.B. Cunningham P.M. Fleming. 1984. 2 & richness (r = 0.65, P < 0.001). Secondly, the New approaches to direct gradient analysis using envi­ life-forms approach enables a considerable ronmental scalars and statistical curve-fitting proce­ dures. Vegetatio 55: 11-27. reduction in the bulk of data ( 18 LP's « 107 spp) without losing ecological information. Austin, M.P., A.O. Nicholls & C.R. Margules, 1990. Thirdly, life-forms represent ecological strate­ Measurement of the realized qualitative niche:environ­ gies per se and, as Cody (1989) states, they con­ mental niches of five Eucalyptus species. Ecol. stitute an index of the number of different ways Monogr. 60: 161-177. that the desert's resources areutilized. Austin, M.P., A.O., Nicholls, M. D. Doherty, & J.A. Meyers, 1994. Determining species response functions ACKNOWLEDGEMENTS to an environmental gradient by means of a B-function. J. Veg. Sci. 5: 215-228.

We thank S. Vargas, J.M. Gallardo, V. Tena and Y. Sandoval for their assistance on the Avery, B.W. & C.L. Bascomb. 1974. Soil survey laborato­ ry methods. Adiar, Harpenden. U.K. field.D. Hodson made valuable suggestions for the improvement of this papero This work was Baker, R.J. & J .A. Nelder. 1978. The GLIM system. supported by the Mexican Ministry of Release 3. Numerical AIgorithms Group, Oxford. Education (SEP-DGICSA; grant NO.C89- Bowers, M.A & c.H. Lowe, 1986. Plant-form gradients on 010216). Sonoran Desert bajadas. Oikos 46: 284-291.

Byers, D.S. 1967. Climate & hidrology, p. 48-65. In Byers, RESUMEN D.S. (ed.) The prehistory of the Tehuacan valley. Environment and subsistence, VoL 1. University of Texas. Austin, Texas. Bajo la consideración de que las características mor­ fológicas representan estrategias conductuales de las plantas para enfrentar presiones ambientales, definimos 18 formas Cody, M.L. 1989. Growth-forms diversity and community de vida utilizando técnicas multivariadas de clasificación en structure in desert plants. J. Arid Environ. 17: 199-209. ZAV ALA-HURTADO et al.: Vegetation-environmentrelationships 589

Dávila, P., J.L. ViIlaseñor, Medina, R. Ramírez,A. Salinas, Shreve, F. 1942. The desert vegetation of North America. A., Sánchez-Ken, J. & Tenorio, P. 1983. Listado Bot. Rev. 8: 195-246. Florístico de México X. Flora del Valle de Tehuacán­ Cuicatlán. Instituto de Biología, UNAM, México, D.F. Smith Ir., C.E. 1965. Flora, Tehuacan valJey. Fieldiana 31: J05-144. Ezcurra, E., M. Equihua & López-Portillo, J. 1987. The desert vegetation of El Pinacate, Sonora, México. Smith Jr., C.E. 1967. Plant remains, p. 220-255. In Byers, Vegetatio 71: 49-60. D.S. (ed.) The prehistory of the Tehuacan valley. Environment and subsistence, Vol. 1. University of Fitter, A.H and R.K.M. Hay, 1983. Environmental physi­ Texas. Austin, Texas. ology of plants. Academic, London. Solbrig, O.T., M.A. Barbour, J. Cross, Goldstein, CH., Gauch, H.G. 1982. Multivariate analysis in community Lowe,CH., Morello, 1. & Yang T.W. 1977. The strate­ ecology. Cambridge University, Cambridge. gies and community patterns of desert plants, p. 67- 107. In Orians, G.H. and O.T. Solbrig, (eds.). Gauch, H.G. & R.H. Whittaker, 1972. Coencline simula­ Convergent evolution in warm deserts: examination of tion. Ecology 53: 446-451. strategies and patterns in deserts of Argentina and the United States. Dowden, Hutchingson Ross. Greig-Smith, P. 1983. Quantitative plant ecology. & Stroudsburg, Pennsylvania. Blackwell Scientific,Oxford. ter Braak, C.J.F. 1985. Correspondence analysis of Jackson, M.L. 1958. Soil chemical analysis. Constable, incidence and abundance data: properties in terms London. of an unimodal response mode!. Biometrics 41: 859-873. Leishman, M. & M. Westoby, 1992. Classifyingplants into groups on the basis of associations of individuals traits. ter Braak, C.J.F. 1987. CANOCO a FORTRAN program evidence from Australian semi-arid woodlands. Eco!. 1. for canonical community ordination by [partial] 80: 417-424. [detrendedJ [canonical] correspondence analysis, prin­ Michin, P.R. 1987. An evaluation of the relativerobustness cipal component analysis and redundancy analysis (ver­ of techniques for ecological ordination. Vegetatio 69: sion 2.1). ITINO. Agriculture Mathematics Group, 89-107. Wageningen, Netherlands. ter Braak, CJ.F. C.W.N. Looman, 1986. Weighted aver­ Montaña, C. 1990. A floristic-structural gradient related to & aging, logistic regression and the Gaussian responce land forms in the southern Chihuahuan Desert. J. Veg. mode!. Vegetatio 65: 3-1 1. Sci. 1: 669-674.

Norusis, M. J./SPSS Ine. 1988. SPSS/PC+ Advanced Villaseñor, J .L., P. Dávila & F. Chiang, 1990. Statistics V 2.0 for the IBM PC/XT/AT and PS2. Fitogeografía del Valle de Tehuacán-Cuicatlán. Statistics Guide. SPSS, Chicago. Boletín de la Sociedad Botánica de México, No. 50: 135-149. Noy-Meir, 1. 1973. Desert ecosystems: environment and producers. Ann. Rev. Eco!. Syst. 4: 25-51. Vite, F., Zava1a-Hurtado, J.A., M.A. Armella, & M.D. García-Suárez,1990. Regionalización y caracterización Noy-Meir, I. 1985. Desert ecosystems structure and func­ macroclimática del Matorral Xerófilo. In Atlas tion, p. 93-103. In Evenari, M.,Noy-Meir, 1. and Nacional de México, Vo!. JI, Carta IV.8.3.B. Instituto Goodall, D.W. (eds.). Hot Deserts and Arid Shrublands. de Geografía, UNAM. México, D.F. Vol. A. Elsevier, Amsterdam.

Okland, R.H. 1992. Studies in SE Fennoscandian mires: Ward, J. 1963. Hierarchical grouping to optimise an obje­ relevance to ecological theory. J. Veg. Sci. 3: 279-284. tive function. J. Amer. Stat. Ass. 58: 236-244.

Raunkiaer, C 1934. The Life Forms of Plants and Statisti­ Zavala-Hurtado, J. A. 1982. Estudios ecológicos en el cal Plant Geography. Oxford University, Oxford. valle semiárido de Zapotitlán, Pue. I. Clasificación numérica de la vegetación basada en atributos binarios Rzedowski, J. 1978. Vegetación de México. Limusa. de presencia o ausencia de las especies. Biótica México. (México) 7: 99-120. 590 REVISTA DE BIOLOGrA TROPICAL

Appendix 1. Plant species used jor fhe definiríon oj life-forms in the semiarid valley oj Zapotitlan, Mexico

Species Life-fonn Speeies Life-fonn

79 Neobuxbaumia tetetzo (Web. ex K. Seh.) Backeb. 1 6 Aeschenomene compacta Rose 11 101 Sfenocereus stellatus (Pffei.) Riccob. 1 58 Lippia graveolens Kunth 11 24 Cephalocereus chrisacanthus (Weber) Britt & Rose 1 40 Echinopteryxeglandulosa Small. 11 77 Neobuxbaumiamacrocephala (Weber) Dawson 1 69 Mascagnia seleriana Loes. 11 76 Myrtil/ocactus geometrízans (e. Martius)Console 1 5 Actinocheita filicina (Sesse & Moe.) Barkley 11 90 chíchipe (Gosselin) Backeb. 1 92 Asteraeeae 7 11 49 Polaskia chende (Gosselin) Gibson & Horak 1 27 Cnidoscolous tehuacanensis Breekon 12 80 Opuntia qecumbens Salm-Dyk 2 36 Croton ciliato-glandulosus Ortega 12 81 Opuntia depressa Rose 2 104 Turnera difussa Willd ex Sehult. 12 83 Opuntia pumita Rose 2 105 Viguiera sp. 12 106 Peniocereus viperina (Weber) Buxb. 2 50 Hibiscus sp. 12 84 Opuntia rosea De Candolle 2 94 Ruellia sp. 12 4 Acantofhamnus aphyllus (Schldl.) Stanley 2 95 Salvia sp. 12 82 Opunfia piliferaWeber 2 23 Ce/tis pallida Torr. 13 39 Echinocactus platyacanthus Link & Otto 3 88 Phitecellobium acaflense Benth. 13 44 Ferocactus recurvus (MilIer) Lindsay 3 51 Ipomoea arborescens Sweet 13 35 Coryphantapalida Britt. & Rose 3 3 Acacia suban[?ulata Rose 13 25 Cephalocereus columna-trajaniSch. 3 75 Morkillia mexicana Rose & Painter 13 78 Neobuxbaumia mezcalaensis (Bravo-Holl.) Backeb. 3 93 Pseudos/1lodin[?iummultifo/ium Rose 13 100 Pachycereus marflinatus(De Candolle) Britt & Rose 3 22 Ceiba parvifoliaRose 13 85 Pachycereus hollianus (Weber) Buxb. 3 99 Senna prin[?lei Rose 13 65 Mammillaria compressa De Candolle " 68 Manihot sp. 13 67 Mammillaria pectinijera (Ruempler) Weber 4 42 Eysenhardtia polystachia Sargo 13 60 Mammi/aria carnea Zuee. ex Pfeiffer 4 2 Acacia sp. 14 62 Mammilaria myxtax Martius 4 74 Mimosa luisana Brandg. 14 66 Mammillaria ele[?ans De Candolle 4 73 Mimosa lacerara Rose 14 43 Ferocactusjlavovirens (Scheidw.) Britt. & Rose 5 91 Prosopis laevi[?afa 45 Ferocacfus robustus (Pfeiffer) Britt. & Rose 5 (Humb. & Bonpl ex WilId.) Johnst. 14 63 Mammilaria sphacelata Martius 5 1 Acacia cOllsfricta Benth 14 64 Mammilaria viperina Purpus 5 26 Cercidiwn praecox Harms 14 61 Mammilaria haageana Pfeiffer 5 19 Caesalpillia melalladenia Stand!. 15 13 Brahea dulcis Cooper 6 37 Da/ea sp. 15 107 Yucca periculosa Baker 6 20 Calliandropsis sp. 15 12 Beaucarneaflraci/is Lemm. 6 34 COI·dia cylindrostachya Roem. & Schult. 15 38 Echeveria sp. 6 54 Karwinskia humboldtiana Zueco 15 7 A[?ave karwinskii Zucc. 7 21 Castela tortuosa Liebm. 15 I1 A[?ave potatorum Zueco 7 72 Megasti[?ma galiotfii Baill 15 9 Agave macrollcanlha Zueco 7 86 Peditanthus aphy/lus Bois, 16 10 A[?ave marmorala Roez!. 7 87 Pedilllnthussp. 16 8 Agave kerchovei Lemm. 7 41 Euphorbia antisiphyllitica Zuee. 16 48 Hechtia podantha Mez. 7 16 Bursera jagaroides Eng!. 17 102 Tillandsia pueblensis Liman Smith 8 17 Bursera hindsiana Engl. 17 103 Tillandsia recurvata L. 8 14 Bursera sp. 17 98 Se/la[?inellasp. 8 18 Burserasp. 17 52 lpomoea conzatii Greenman 9 15 Bursera arida Stand!. 17 71 Metaste/ma sp. 9 59 Lycium sp. 18 70 Maximowitzia sp. 9 97 Schaefferia stenophyllaStand!. 18 57 Lippia gratissima (GilI.) Troncoso 10 53 Jalropha spalhulata Muell. Arg. 18 96 Sallvitaliafruticosa Hemsley 10 47 Gymnosperma [?lutinosaSpreng. 18 55 Lanlana camara L. 10 89 Plumeria rubraL. 18 56 Lanfana sp. 10 46 Fouquieriaformosa Kunth 18 32 Asteraceae 1 10 33 Asteraecae 2 10 28 Asteraeeae 3 10 30 Asteraceae 4 10 31 Asteraceae 5 10 29 Asteraeeae 6 10