Morphoanatomical leaf traits of herbaceous species raised in a cronosequence of naturally regenerated areas of tropical deciduous forest

ABSTRACT

Plant survival in semi-arid climates depends on morphological, anatomical, and physiological adjustments that determine survival under limiting environmental conditions. In this study, anatomical and morphological characteristics of three herbaceous species with different growth habits were determined: a tall herb, Mesosphaerum suaveolens (tall), Stachytarpheta sanguinea (short), and Jacquemontia evolvuloides (climbing). Fully expanded leaves were collected from three sites in each of three different natural regeneration periods in a seasonally dry tropical forest. Leaf, specific leaf area, epidermis, cuticle, parenchyma, and mesophyll thicknesses; stomatal and trichome densities were measured. All three species have stomata and trichomes on both leaf surfaces and higher densities on the abaxial surface than the adaxial surface. M. suaveolens and S. sanguinea have unusual leaf morphological and anatomical adjustments, with a larger leaf area in the more open and youngest site, but thicker abaxial and adaxial epidermis and mesophyll in the oldest site which has the closest canopy stratum. J. evolvuloides has more usual adjustments, with a smaller specific leaf area and thinner epidermis and isobilateral parenchyma in the youngest site, being better acclimatized to low water availability and high light incidence. Therefore, M. suaveolens and S. sanguinea have unique acclimatization strategies for different environmental conditions.

KEYWORDS:

• Caatinga
• ecological succession
• herbaceous plant strategies
• leaf morphology; leaf anatomy
• plant community

Introduction

Plant survival under different environmental conditions depends on their morphological, anatomical, and physiological adaptations [1,2]. Leaves in arid and semi-arid climates may adapt to water shortages by developing morphological traits such as reduced surface/volume ratios, thickened cuticles, external periclinal walls of epidermal cells, high trichome and stomatal densities, elongated substomatal chambers, well-developed palisade parenchyma, and idioblasts with crystals [1–5]. The adjustments to water and light availability vary in plants with different habits. For instance, small herbs may have narrower leaves, with thinner adaxial epidermis, more trichomes on both surfaces and lower specific leaf areas than vines [6]. The adjustments may also change with the succession stage in the native vegetation undergoing regeneration processes, which depend on the local environmental conditions.

The Brazilian tropical deciduous forest, known as caatinga [7], is one of the largest and driest tropical forest formations in the world [8,9], and most herbaceous plants within it have relatively brief life cycles restricted to the short rainy seasons. Multiple small plots of native vegetation in the Brazilian tropical deciduous forest are repeatedly deforested to provide fuelwood or to be planted with crops and pastures. Many of these plots are abandoned, in the prevailing itinerant agricultural system, and the native caatinga vegetation naturally regenerates [10,11]. Grasses, herbaceous plants, and young shrubs and trees compete in the early regeneration stages [12]. As the native shrubs and trees develop and the planted grasses disappear, the herbaceous plants become shaded, and they concentrate in better-illuminated canopy interval spaces [12,13]. This process may create different morphoanatomical adaptations which vary with plant growth habit and light exposure [14,15]. Nevertheless, there is little information about the morphoanatomical adaptations of herbaceous plants to water and luminosity conditions in this and in other semi-arid regions of the world.

The objective of this work was to examine the adjustments and morphological and anatomical characteristics of three herbaceous species with contrasting growth habits in caatinga sites with different regeneration periods: a tall herb, Mesosphaerum suaveolens (L.) Kuntze (Lamiaceae); a small herb, Stachytarpheta sanguinea Mart. ex Schauer (Verbenaceae); and a climbing herb, Jacquemontia evolvuloides (Moric.) Meisn (Convolvulaceae). The hypothesis is that the different growth habits of the three herbaceous species, resulting in different exposures to sunlight, influence their foliar morphology and anatomy in different ways along the successional stages, with differences in leaf area, stomata and trichome density, and epidermis and mesophyll thickness.

Materials and methods

Study area and species

This study was carried out in naturally regenerated areas of caatinga at the Fazenda Tamanduá (7°01′31′′S and 37°23′31.8′′W) in Santa Teresinha municipality, Paraíba state, Brazil. The climate of the region is Bsh (semi-arid) according to the Köppen classification system [16]. The average annual temperature is 28°C and the average annual rainfall is 800 mm, concentrated in 3 to 4 months. The soil is classified as a Litholic Neosol [17].

Three caatinga sites were selected for each of the three different regeneration periods [18]: 1) three sites which were deforested, then planted with perennial cotton and later with Cenchrus ciliaris L. (Poaceae) grass and where the native vegetation had been naturally regenerating when the pasture use was discontinued in 1992. Therefore, the sites had been regenerating for 22 years and they were designated R22. the legume trees (Mimosa tenuiflora (Wild.) Poir, dominate these sites, forming more than 90% of the total canopy cover of 6045 m² ha⁻¹; 2) a second group of three sites which were planted with cotton from 1965 to 1970, were undergoing natural regeneration for 44 years and, thus, were designated R44. Their average canopy covers were 6278 m² ha⁻¹; and 3) a third group of three sites, which had been covered with caatinga vegetation, uncut and without major perturbations since at least 1950, with a canopy cover of 8458 m² ha⁻¹. Since the exact regeneration period was unknown, but knowing it was more than 60 years, the sites were designated R > 60. The distance from one site to any other of any regeneration period was less than 3 km [18].

One 60 m × 30 m plot was delimited at each site during each regeneration period. The three most abundant herbaceous species in all plots, previously determined in a phyto-sociological survey [19], were selected for our study. They were Mesosphaerum suaveolens (L.) Kuntze, Stachytarpheta sanguinea Mart. ex Schauer, and Jacquemontia evolvuloides (Moric.) Meisn.

M. suaveolens is a tall herb species, averaging 1.5 m height when fully grown, grows fast, and can reach 3 m in height [20]. It is often found in areas which have undergone natural or human disturbance [21]. It is considered an invasive species and has been the target of biological control programs in Australia [22]. S. sanguinea has a low stature, averaging 0.4 m height, and it is endemic to Brazil [23]. It grows up to 40 cm in the experimental area, preferentially in high light intensity spots. J. evolvuloides is a climbing species, native of Brazil [24], which may cover the soil and inhibits the growth of other plant species [25]. It grows over other herbaceous plants in the study area, but it did not cover the taller M. suaveolens plants. All three species (as most other herbs in the area) only grow during the rainy season. M. suaveolens has a six-month life cycle, while the other two species have four-month cycles. All three species have fasciculated roots without tubers.

Collection and data analysis

Five fully expanded leaves were removed during the rainy season with pruning shears from 10 individuals of each species at each site for anatomical and morphological analysis. Each individual was randomly selected within a different quadrat of consecutive quadrats that were randomly distributed within each plot, until 10 of the quadrats had at least one plant of the selected species. The full expansion of the leaves was ascertained by a combination of criteria, including color, shape, size, and distance from the stem or branch apex (fifth distal node). On the same day of the leaf collection, measurements were made of soil moisture, vapor pressure deficits, and photosynthetic photon flux densities in each plot (Table 1) to indicate differences in microclimatic variables, possibly caused by differences in vegetation, mainly canopy cover.

Table 1. Average soil moisture, vapor pressure deficits, and photosynthetic photon flux densities in sites with different regeneration periods (22 years, R22; 44 years, R44; and more than 60 years, R > 60) after land abandonment, in a northeastern Brazilian seasonally dry tropical forest (Santa Teresinha, Paraíba state, Brazil).

	R22	R44	R >60
Soil moisture (%)	6.66	9.33	9.00
Vapor pressure deficits (kPa)	2.96	1.15	1.52
Photosynthetic photon flux densities (µmol m^−2 s^−1)	1600	1300	1500

The leaves were scanned to measure their surface area (cm²) and maximum length (cm) and width (cm). All measurements were performed using the digital image analysis program ImageJR software ver. 1.45a (NIH, Bethesda, MD). The leaves were then oven-dried at 70°C and weighed. The specific leaf area (SLA) was calculated as the ratio between the leaf area and dry weight [26]. The description and classification of the leaf morphological structure were based on the terminology used by Barroso et al. [27].

Three fully expanded leaves, following the same criteria previously described, were also removed, basipetally between the third- and fifth nodes, from each of three plants per species per site, for anatomical analyses. The leaves were fixed in FAA50 (formaldehyde, acetic acid, and 50% ethanol) for 48 h, then placed in 70% ethanol, and dehydrated in an alcohol series ranging from 50% ethanol/50% butanol to 100% butanol [28]. Median leaf blade fragments (excluding the midrib) were embedded in paraffin, pressed, and transversely sectioned with a rotating microtome (Zeiss, Model Hyrax M 55) at thicknesses ranging from 10 to 12 μm. The sections were stained with Alcian blue and safranin [29] and 2 mm² samples were taken to perform anatomical analysis. Twenty micrometer measurements were made of the mesophyll, adaxial and abaxial epidermis, cuticle and palisade, and spongy parenchyma thickness of each individual. The sections were stained with Alcian blue and safranin [29]. Samples of 2 mm² were collected from the median region of the leaves to perform anatomical sections. From these sections, leaf blades were mounted in Canada Balsam. We have taken thickness measurement of the mesophyll, adaxial and abaxial epidermis, cuticle and palisade, and spongy parenchyma using a micrometer. We did twentymeasurements per individual (micrometer) of the mesophyll, adaxial and abaxial epidermis, cuticle and palisade, and spongy parenchyma were taken.

Stomatal and trichome densities (per mm²) were determined after separating the epidermis, following the technique developed by the authors: the epidermis of the leaf blade fragments was dissociated under a magnifying glass using deionized water, and a soft brush and the dissociated epidermal fragments were washed in distilled water, stained with safranin, and mounted on 50% glycerin [30]. The slides were sealed with colorless nail polish and observed under a photomicroscope. Trichome sizes and stomatal and trichome densities in 1 mm² areas were measured using the ImageJR software ver. 1.45a (NIH, Bethesda, MD). The stomata and trichomes were classified according to the method of Metcalfe and Chalk [31].

The data were analyzed with ANOVA, after verifying that they were normally distributed and homogeneous, using the Shapiro-Wilk and Levene tests, respectively. A different ANOVA was run for each leaf characteristic and each species; thus, comparisons were made of the means of the regeneration periods for each species and each leaf characteristic. Data from each of the three sites in each regeneration period were used as replicates. The comparisons were made with the Tukey test at a 5% probability level [32]. The Statistica 7.0 software (StatSoft, 2004) was used for all statistical analyses [33].

Results

Mesosphaerum suaveolens leaves had oval lamina, acute apices, obtuse bases, and serrated margins (Figure 1A). Stachytarpheta sanguinea leaves had linear, narrow, and long lamina, sharp apices, truncated bases, and serrated margins (Figure 1B). Jacquemontia evolvuloides leaves had orbicular lamina, obtuse apices, cordiform bases, and entire (smooth) margins (Figure 1C).

Figure 1. Morphological characteristics of the leaves of Mesosphaerum suaveolens (A), Stachytarpheta sanguinea (B) and Jacquemontia evolvuloides (C) harvested from individuals growing in sites with different regeneration periods after land abandonment in a northeastern Brazilian seasonally dry tropical forest (Santa Teresinha, Paraíba state, Brazil).

Only the maximum leaf length in M. suaveolens differed significantly between plants growing in the sites of different regeneration periods (Table 2; F_{(4, 142)} = 9.40, P < 0.01). It was the highest in plants growing in the sites regenerating for 44 years (R44) and lowest in the sites regenerating for 22 years (R22). The maximum leaf length (F_{(4, 142)} = 6.81, P < 0.01), leaf area (F_{(4, 142)} = 7.68, P < 0.01), and maximum leaf width (F_{(4, 142)} = 6.81, P < 0.01) in S. sanguinea were all highest in R22, whereas the specific leaf area was highest in R44 (F_{(4, 142)} = 7.68, p < 0.01). J. evolvuloides leaves only differed in terms of specific leaf area (SLA), which was highest in R > 60 (F_{(4, 142)} = 6.83, p < 0.01).

Table 2. Morphological characteristics in leaves of three herbaceous species (one tall, one short, and one climbing species) growing in forest sites with different regeneration periods after land abandonment in a northeastern Brazilian seasonally dry tropical forest (Santa Teresinha, Paraíba state, Brazil).

Species	Regeneration periods
	R22	R44	R >60
Leaf area (cm2)
Mesosphaerum suaveolens	11.9 ± 3.8^a	12.6 ± 5.6^a	11.9 ± 3.3^a
Stachytarpheta sanguinea	9.45 ± 3.24^a	6.97 ± 1.77^b	5.9 ± 1.3^b
Jacquemontia evolvuloides	6.83 ± 2.18^a	7.38 ± 1.73^a	7.97 ± 2.58^a
Specific leaf area (cm^2 g^−1)
Mesosphaerum suaveolens	196.3 ± 20.3^a	164.02 ± 27.7^a	156.2 ± 27.2^a
Stachytarpheta sanguinea	304.1 ± 41.7^b	383.2 ± 54.4^a	231.27 ± 28.9^c
Jacquemontia evolvuloides	87.1 ± 10.8^c	191.6 ± 18.01^b	240.61 ± 26.2^a
Leaf blade length (cm)
Mesosphaerum suaveolens	4.49 ± 0.86^b	5.38 ± 0.99^a	4.98 ± 0.70^ab
Stachytarpheta sanguinea	5.38 ± 1.04^a	4.65 ± 0.64^b	4.250 ± 0.47^b
Jacquemontia evolvuloides	3.55 ± 0.62^a	3.66 ± 0.44^a	3.38 ± 0.51^a
Leaf blade width (cm)
Mesosphaerum suaveolens	3.41 ± 0.62^a	3.67 ± 0.73^a	3.59 ± 0.55^a
Stachytarpheta sanguinea	2.38 ± 0.49^a	2.03 ± 0.33^b	1.87 ± 0.26^b
Jacquemontia evolvuloides	2.68 ± 0.48^a	2.752 ± 0.38^a	2.82 ± 0.54^a

Averages in the column followed by the same letter are not significantly different by the Tukey test at 5% probability level (n = 50 ± standard deviation). R22 = site under regeneration for 22 years; R44 = site under regeneration for 44 years; R > 60 = site under regeneration for over 60 years.

All three species had amphistomatic leaves. M. suaveolens and S. sanguinea had diacytic stomata, whereas those of J. evolvuloides were paracytic (Figures 2a, 3 a and 4a). The abaxial surfaces of M. suaveolens leaves in R > 60 had the highest stomatal density of all (F_{(1, 114)} = 187.32, p < 0.01). Only the adaxial stomatal densities in S. sanguinea (F_{(1, 114)} = 9.82, p < 0.01) and J. evolvuloides (F_{(1, 114)} = 105.48, p < 0.01) significantly differed between successional stages (Table 3).

Figure 2. Mesosphaerum suaveolens harvested from individuals growing in sites with different regeneration periods after land abandonment in a northeastern Brazilian seasonally dry tropical forest (Santa Teresinha, Paraíba state, Brazil). Diacytic stomata; tector and glandular trichomes in frontal view (A-C); cross-section of leaf blade (D-E) and midrib (F). (D) GT = trichomes; (E) PP = palisade parenchyma; LP = lacunar parenchyma; ep aba = abaxial epidermis; ep ada = adaxial epidermis; (F) VB = vascular bundle; ep = epidermis; FP = fundamental parenchyma; Phl= phloem; xy = xylem; col = collenchyma;; sto = stomata.

Figure 3. Stachytarpheta sanguinea harvested from individuals growing in sites with different regeneration periods after land abandonment in a northeastern Brazilian seasonally dry tropical forest (Santa Teresinha, Paraíba state, Brazil). Diacytic stomata on the adaxial leaf surface; hook-shaped tector trichomes; multicellular tector trichomes; tector trichomes at high density in the vein (A-D) in frontal view; cross-section of leaf blade (E) and midrib (F). (E) PP = palisade parenchyma; LP = lacunar parenchyma; TT = tector trichomes; sto = stomata; ep aba = abaxial epidermis; ep ada = adaxial epidermis; (F) ep = epidermis; FP = fundamental parenchyma; phl = phloem; xy = xylem; col = collenchyma; TT = tector trichomes.

Figure 4. Jacquemontia evolvuloides leaves harvested from individuals growing in sites with different regeneration periods after land abandonment in a northeastern Brazilian seasonally dry tropical forest (Santa Teresinha, Paraíba state, Brazil). Paracytic stomata and star-shaped tector trichomes (A-B) in frontal view; cross-section of leaf blade (C) and midrib (D). (C) PP = palisade parenchyma; LP = lacunar parenchyma; T = trichome; ep aba = abaxial epidermis; ep ada = adaxial epidermis; (D) ep = epidermis; FP = fundamental parenchyma; phl = phloem; xy = xylem.

Table 3. Morphometric characteristics in leaves of three herbaceous species (one tall, one short, and one climbing species) growing in sites with different regeneration periods after land abandonment in a northeastern Brazilian seasonally dry tropical forest (Santa Teresinha, Paraíba state, Brazil).

Species	Leaf Surface	Regeneration periods
		R22	R44	R >60
Epidermis thickness (μm)
Mesosphaerum suaveolens	Abaxial	6.5 ± 1.4^Aa	7.2 ± 2.35^Aa	6.3 ± 1.73^Ab
	Adaxial	6.9 ± 1.7^Aa	7.75 ± 2.14^Aa	7.7 ± 2.53^Aa
Stachytarpheta sanguinea	Abaxial	6.72 ± 1.4^Ba	7.0 ± 1.8^Ba	8.2 ± 1.3^Aa
	Adaxial	6.92 ± 1.44^Ba	7.3 ± 1.6^Ba	8.13 ± 2.08^Aa
Jacquemontia evolvuloides	Abaxial	10.1 ± 2.86^Aa	10.7 ± 2.1^Aa	11.64 ± 2.4^Aa
	Adaxial	10.9 ± 2.48^Aa	11.2 ± 2.6^Aa	11.28 ± 1.7^Aa
Cuticle thickness (μm)
Mesosphaerum suaveolens	Abaxial	1.3 ± 0.4^Aa	1.2 ± 0.5^Aa	1.1 ± 0.3^Aa
	Adaxial	1.3 ± 0.4^Aa	1.3 ± 0.4^Aa	1.2 ± 0.4^Aa
Stachytarpheta sanguinea	Abaxial	1.1 ± 0.3^Aa	1.1 ± 0.3^Aa	1.0 ± 0.0^Aa
	Adaxial	1.3 ± 0.3^Aa	1.2 ± 0.4^Aa	1.03 ± 0.1^Ba
Jacquemontia evolvuloides	Abaxial	1.2 ± 0.4^Aa	1.1 ± 0.3^Aa	1.1 ± 0.3^Aa
	Adaxial	1.3 ± 0.5^Aa	1.0 ± 0.0^Aa	1.3 ± 0.4^Aa
Stomatal density (n mm−2)
Mesosphaerum suaveolens	Abaxial	275.1 ± 73.1^ABa	247.8 ± 41.0^Ba	301.8 ± 54.1^Aa
	Adaxial	183.6 ± 25.0^Ab	135.7 ± 15.5^Bb	183.0 ± 40.4^Ab
Stachytarpheta sanguinea	Abaxial	175.7 ± 35.0^Aa	196.9 ± 32.3^Aa	195.1 ± 34.9^Aa
	Adaxial	154.5 ± 33.1^Ba	155.7 ± 24.6^Bb	192.7 ± 57.9^Aa
Jacquemontia evolvuloides	Abaxial	105.4 ± 18.9^Aa	123.0 ± 23.0^Aa	124.84 ± 32.4^Aa
	Adaxial	100.0 ± 20.7^Aa	76.3 ± 13.1^Ab	56.96 ± 12.4^Bb
Trichome density (n mm−2)
Mesosphaerum suaveolens	Abaxial	5.0 ± 7.9^Aa	0.3 ± 0.2^Ba	1.1 ± 0.4^Ba
	Adaxial	1.14 ± 0.6^Ab	1.1 ± 0.3^Aa	1.7 ± 0.5^Aa
Stachytarpheta sanguinea	Abaxial	12.1 ± 8.3^Aa	14.28 ± 15.1^Aa	15.1 ± 3.48^Aa
	Adaxial	0.8 ± 0.4^Bb	1.6 ± 0.5^Bb	7.3 ± 3.3^Ab
Jacquemontia evolvuloides	Abaxial	3.0 ± 0.7^Ba	3.8 ± 1.2^Ba	7.4 ± 1.6^Aa
	Adaxial	2.7 ± 1.0^Aa	2.5 ± 0.6^Ab	3.4 ± 0.5^Ab
Trichome size (μm)
Mesosphaerum suaveolens	Abaxial	58.8 ± 10.8^Aa	43.4 ± 5.02^Ba	44.4 ± 6.5^Ba
	Adaxial	48.0 ± 7.38^Aa	56.2 ± 6.2^Aa	52.4 ± 7.9^Aa
Stachytarpheta sanguinea	Abaxial	31.9 ± 23.6^Aa	28.4 ± 18.0^Aa	26.7 ± 19.8^Ab
	Adaxial	35.0 ± 26.4^Aa	33.4 ± 25.2^Aa	45.3 ± 34.0^Aa
Jacquemontia evolvuloides	Abaxial	22.6 ± 2.07^Aa	23.4 ± 2.7^Aa	16.8 ± 4.7^Ab
	Adaxial	27.2 ± 3.7^Aa	33.6 ± 7.09^Aa	33.8 ± 8.1^Aa
Mesophyll (µm)
Mesosphaerum suaveolens		78.1 ± 17.6^B	89.6 ± 9.3^A	93.8 ± 5.5^A
Stachytarpheta sanguinea		79.8 ± 8.7^B	69.9 ± 12.4^C	89.9 ± 16.15^A
Jacquemontia evolvuloides		74.2 ± 8.1^A	65.2 ± 11.8^B	81.3 ± 7.7^A
Palisade parenchyma (μm)
Mesosphaerum suaveolens		31.4 ± 10.3^A	32.5 ± 4.1^A	35.6 ± 5.1^A
Stachytarpheta sanguinea		31.16 ± 4.57^A	26.8 ± 7.07^B	30.5 ± 7.02^AB
Jacquemontia evolvuloides	Abaxial	12.1 ± 2.0^Ab	11.2 ± 3.6^Bb	14.2 ± 3.3^Ab
	Adaxial	23.5 ± 3.9^Aa	19.1 ± 5.5^Ba	27.0 ± 2.6^Aa
Lacunar parenchyma (μm)
Mesosphaerum suaveolens		33.2 ± 8.8^B	43.8 ± 5.7^A	43.6 ± 4.9^A
Stachytarpheta sanguinea		35.5 ± 6.4^B	28.8 ± 5.7^C	42.9 ± 11.8^A
Jacquemontia evolvuloides		13.6 ± 3.1^AB	12.6 ± 2.8^B	16.0 ± 3.4^A

Averages (± standard deviation) in the column followed by the same capital letter for species comparisons and by the same lower case letter for surface comparisons are not significantly different by the Tukey test at 5% probability level (n = 90). R22 = site under regeneration for 22 years; R44 = site under regeneration for 44 years; R > 60 = site under regeneration for over 60 years.

All three species had trichomes on both leaf surfaces. The trichomes in M. suaveolens had the highest density and the largest size on the abaxial leaf surfaces in plants growing in R22 (F_{(2, 24)} = 6.79, P < 0.01). The highest trichome density in J. evolvuloides occurred on the abaxial leaf surfaces of plants in R > 60 (F_{(1, 114)} = 92.47, P < 0.01), while the trichome densities in S. sanguinea were greater on the abaxial than adaxial leaf surfaces in all sites (F_{(1, 114)} = 186.07, P < 0.01). The trichome size in these two species was greater on the adaxial than abaxial leaf surface, but only in R > 60 plants (Table 3). All three species had pluricellular tector trichomes, with those of M. suaveolens being glandular-sited, while those of S. sanguinea were simple hook-shaped, and those of J. evolvuloides were star-shaped (Figures 2c, 3 b and 4b).

The epidermises of the three species in cross-sections were unstratified and covered by a thin cuticle (Fig. 2e, 3 e and 4c). The adaxial leaf epidermis of M. suaveolens in R > 60 was thicker than that on the abaxial surface (F_{(10, 146)} = 4.85, P < 0.01). Both the abaxial and adaxial leaf epidermises of S. sanguinea were significantly thicker in R > 60 than in the other successional stages (F_{(2, 164)} = 10.71, P < 0.01) (Table 3).

The mesophylls in M. suaveolens and S. sanguinea were dorsiventral and consisted of a cross-sectionally unstratified palisade parenchyma facing the adaxial surface. The leaf mesophyll in J. evolvuloides was isobilateral and composed of two palisade parenchyma cell layers on the adaxial side and one on the abaxial surface. All three species had multi-stratified lacunar parenchyma. The lacunar parenchyma in M. suaveolens consisted of 4–5 cell layers, with sparse druses in plants of R22 and R44. Moreover, the lacunar parenchyma in S. sanguinea had 5–6 layers, while there were 2–3 layers loosely arranged in an irregular distribution in J. evolvuloides.

The leaf mesophyll of M. suaveolens was thicker in R44 and R > 60 plants than in R22 plants (F_{(10, 146)} = 4.85, P < 0.01). The mesophyll in S. sanguinea was thicker in R > 60 than in the other two successional stages (F_{(6, 160)} = 10.47, P < 0.01). In addition, the mesophyll in J. evolvuloides was thicker in R22 and R > 60 than in R44 (F_{(4, 80)} = 5.82, P < 0.01).

The midribs of all three species were biconvex. They contained fundamental parenchyma composed of rounded cells. The vascular bundle was collateral. The xylem occupied the central region of the vein and formed an arch with the opening facing the adaxial leaf surface. The phloem was in the external region of the xylem. All three species had multiseriate angular collenchymata. This tissue had five cell layers in M. suaveolens, four in S. sanguinea, and three in J. evolvuloides (Figure 4F).

Discussion

Several morphological and anatomical characteristics of the three species were different among the three regeneration periods, demonstrating that the species developed acclimations to the different microclimatic conditions generated by modifications in the vegetation along the successional stages. These modifications involve mainly increasing shading and decreasing light penetration within the canopy but may also involve increasing evapotranspiration and decreasing water availability (Table 1). The different growth habits of the species influence their adaptations, mainly to light availability, the climbing and the tall herb species probably receiving more radiation than the small herb species. However, another form of adaptation is the preferential growth of each species in microsites within the vegetation. The climbing species may grow near trees and shrubs, while the small herb species may grow in open spots. These preferences were not systematically measured, but field observations were registered and may help explain their morphological and anatomical acclimations.

The external leaf morphology of M. suaveolens confirms that described by Lorenzi and Matos [21] and Zoghbi et al. [34], except for the blade base. The leaves of the plants at Santa Teresinha were obtuse rather than cordiform. The leaf morphology of S. sanguinea differed from those previously reported. Santos et al. [35] stated that Stachytarpheta microphylla Walp. (synonyms of S. sanguinea) had ovate blades, acute bases, serrated margins, and acute apices. Only the last characteristic was confirmed in the plants in Santa Teresinha. J. evolvuloides had orbicular blades, obtuse apices, and cordiform bases. In contrast, the same species growing in the Cerrado region of Central Brazil had oval blades, oblong or elliptical shapes, acute apices, and rounded bases [25]. Differences in water availability between the Cerrado and the Caatinga, in the Central and Northeast Brazilian regions, respectively, may account for the discrepancies in leaf morphology between plants of the same species.

Basílio et al. [36] described similar M. suaveolens leaf dimensions to those found in this study (4.0–8.0 cm × 3.0–5.5 cm vs. 11.9–12.6 cm × 3.4–3.6 cm) [37]. The lack of differences in the leaf area and specific leaf area of M. suaveolens among the different regeneration [38] periods may indicate that the shape of the leaf is a genetically determined character that is little affected by the different environmental conditions where the species grows [39–41]. However, it must be reminded that the macro-environmental conditions are the same in all regeneration periods and the differences refer to microclimatic modifications generated by vegetation changes along succession.

The largest leaf areas of S. sanguinea were found in the youngest sites (R22), where the cover of arboreal and shrub canopies is more open (60%) than in the older sites (63–85%). This suggests that this species developed an adaptation mechanism to areas with higher light incidence, since it is known that some species invest in larger leaf areas in sunnier environments [41]. The water conservation strategy in S. sanguinea may be a reduction in abaxial leaf stomatal density rather than leaf area.

J. evolvuloides plants had the lowest specific leaf area in R22 because they had the lowest leaf mass in this area. These plants probably received more light in this area (Table 2) because the area is the one with the more open canopy and because they are vines which can climb to high positions in the canopy. In general, dry matter production varies directly with light incidence [42].

High stomatal density is commonly observed in the abaxial surface of leaves of plants growing in sunny sites, the surface opposed to the light source [34,43,44]. This has been considered an adaptation to minimize water loss [44,45]. However, higher stomatal densities occurred in the leaves of M. suaveolens and S. sanguinea plants growing at the site with the longer regeneration period, which has the highest canopy cover (85%) and, thus, the lowest average light penetration [46]. The possible cause of this apparent unexpected result is that the plants are not randomly dispersed under the canopy but grow selectively in spots where sunlight penetrates the canopy and illuminates them. Neither stomatal density nor distribution had been previously described for S. sanguinea leaves. Nevertheless, it has been reported that several species of Verbenaceae (Lippia turbinate f. magnifolia Moldenke [47]; Bouchea fluminensis (Vell.) Mold [48]; Lantana radula Swartz and Lantana camara L. Voucher [47]; Priva lappulacea L [49,50]) have amphistomatic leaves [36,51]. According to Greulach [52], most of their stomata are on the abaxial surface. There are also no existing reports on the leaf stomatal distribution in J. evolvuloides, but Operculina macrocarpa (Linn.) Urb [44], which belongs to the same family, has stomata on both leaf surfaces.

M. suaveolens had the longest and densest leaf trichomes in R22, with the lowest canopy cover, and highest light incidence. This morphological adaptation may increase the reflection of solar radiation, helping to maintain low leaf temperatures and retention of water vapor around the leaf [1, 54–56]. The abaxial leaf surface in both S. sanguinea and J. evolvuloides had the highest trichome densities, whereas the adaxial surface had the longest trichomes. The high density and large size of S. sanguinea and J. evolvuloides leaf trichomes in the older caatinga were not expected. The greatest trichome density and size tend to occur in areas with some degree of stress, such as high luminosity and/or nutrient- and water deficiencies [53,54]. Martínez-Natarén et al. [55] found the lowest number of trichomes in Lippia graveolens Kunth (Verbenaceae) leaves growing in arid environments. Low soil water and nutrient availability could limit photosynthetic production to a level which would reduce biomass formation, including trichomes. Therefore, plants develop other stress response mechanisms such as enhanced stomatal closure efficiency. This strategy is important for short-cycle caatinga plants, but these results contradict the responses observed elsewhere in most species, including M. suaveolens. Therefore, phenotypic modifications may be species-specific, only occur under certain environmental conditions, and cannot be generalized.

The relatively greater thickness of unstratified epidermis on the adaxial leaf surfaces of M. suaveolens was described by Basílio et al. [37] and Silva [56]. This trait was observed in M. suaveolens growing in the oldest regeneration sites of Santa Teresinha and probably developed because more light reaches the adaxial than the abaxial leaf surfaces. The epidermal cells on the adaxial surface were larger than those on the abaxial surface as an acclimation response to a higher luminosity [57].

The layers in S. sanguinea leaf epidermis had not been described before, but other species of the same family, such as Lippia sidoides, also have unstratified epidermis [57]. Both abaxial and adaxial leaf epidermises were thicker in the leaves of plants growing in R > 60. This is an unexpected result considering that this site has the greatest tree and shrub canopy cover and the epidermis tends to be thicker in sunnier environments. One explanation could be that these plants grow below the largest gaps in the tree canopy, where light penetrates.

There have been no previous reports of unstratified epidermis in J. evolvuloides, but there are records of this trait in other species of the same family, such as Merremia tomentosa (Choisy) Hallier f [58]. J. evolvuloides had the greatest epidermal thickness among the three species. J. evolvuloides is a vine, so it can grow above the other herbaceous species and receive more light. Castro [59] stated that plants grown in full sunlight have thicker epidermises than those grown in the shade. The climbing habit of J. evolvuloides might also account for the fact that there were no significant differences in epidermal thickness among the sites with different vegetation regeneration periods. Alternatively, there may be relatively little plasticity of this trait in J. evolvuloides.

The dorsiventral mesophyll of M. suaveolens was described by Maia [43]. There is no description of the mesophyll type of S. sanguinea, but there are records of species of the Verbenaceae family which also have dorsiventral mesophyll [51,52,60].

The isobilateral mesophyll of J. evolvuloides is an important caatinga acclimatization characteristic and has already been described for M. tomentosa (Choisy) Hall [61]. (Convolvulaceae) by Pereira [58]. This trait enables these plants to survive in semi-arid climates [62]. There is palisade parenchyma on both leaf surfaces of J. evolvuloides, and it fills most of the mesophyll, leaving few gaps. This structure indicates that the plant is highly active photosynthetically because it receives strong solar radiation. This type of leaf anatomy is characteristic of plants exposed to high luminous intensities [1,63].

Mesophyll, palisade, and lacunar parenchyma are generally thicker in plants exposed to higher light incidence [60]. However, the mesophyll in M. suaveolens was relatively thicker in the oldest successional stage whose canopy was dense with arboreal and shrubby plants and where the herbaceous density was lower than those of the younger regeneration stages. S. sanguinea also had a relatively thicker mesophyll in the oldest successional stage. One possible explanation is the preferential growth of the plants of this species below the gaps in the tree canopies, where there is high light incidence.

The relatively higher stomatal densities on both leaf surfaces may compensate for the relatively greater mesophyll thickness of the plants in the oldest successional stage. According to Boeger and Gluzezak [64], greater leaf thickness limits internal CO₂ diffusion because the gases must move relatively greater distances inside the mesophyll. The presence of stomata on both leaf surfaces effectively shortens this distance and optimizes the foliar CO₂ conductance [65,66].

The calcium oxalate druses in the lacunar parenchyma of M. suaveolens in the two youngest successional stages may be related to a greater tissue support and may reduce the digestibility of the leaves and protect them against herbivory [62,67]. In addition, crystals in plants sequester excess calcium carried by the transpiration water flux and xylem sap loss [68] and protect them against UV rays [1].

Conclusions

The three species have different morphological and anatomical adjustments to the progressive successional stages. Most of these adjustments require little energetic expenditure, such as changes in size and density of stomata and trichomes, probably because of the short life cycle of the species and the harsh environment. The adjustments followed the expected patterns in the two younger successional stages, but all three species have unexpected characteristics in the oldest successional stage, such as thicker mesophyll and palisade parenchyma. The preferential growth of the plants in the spots receiving more light, under more open canopy spaces, may explain part of these apparent unexpected characteristics.

M. suaveolens and S. sanguinea have unusual morphoanatomical characteristics, with larger leaf area in the more open earlier successional stage, but thicker abaxial and adaxial epidermis and mesophyll in the oldest successional stage, which has the closest canopy stratum. On the other hand, J. evolvuloides, a climbing species, has more usual adjustments, with smaller specific leaf area and thinner epidermis and isobilateral parenchyma in the earlier successional stage, being better acclimatized to low water availability and high light incidence. Therefore, the growth habit of each species determines their different adjustments to the different microclimatic characteristics found along the successional stages of this very dry seasonally deciduous tropical forest.

Acknowledgments

This work was supported by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (Sisbiota CNPq - 563304/2010-3), and the Fundação de Amparo à Pesquisa de Minas Gerais (FAPEMIG CRA - APQ-00001-11). We gratefully acknowledge Instituto Fazenda Tamanduá for logistical support and Mr Pierre Landolt for allowing us to develop research at Fazenda Tamanduá. The first author thanks the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brasil (Finance Code 001) for her doctorate scholarship. The author J.S. Almeida thanks CNPq (PQ- 307422/2012-7 PQ 309965/2016-0) for scientific productivity fellowship; to Dárdano de Andrade Lima (IPA) and UFP – Geraldo Maris (UFPE) herbariums.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work was supported by the Conselho Nacional de Desenvolvimento Científico e Tecnológico [PQ 309965/2016-0]; Conselho Nacional de Desenvolvimento Científico e Tecnológico [PQ- 307422/2012-7]; Conselho Nacional de Desenvolvimento Científico e Tecnológico [Sisbiota CNPq - 563304/2010-3].