ADIRECTIONALCLINE IN MOURIRI GUIANENSIS (ME LASTOM AT ACE AE) Thomas Morley ( ;t) Abstract of specialization of the most important variable, the ovary, can be clearly identi­ Morphological variation in Mouriri guia- fied. The overall pattern of distribution nensis is described and analyzed throughout its range in Brazil and adjacent regions. Featu­ was briefly described previously (Morley, res that vary are ovary size, locule and ovule 1975, 1976); the present paper is a detai­ number, shape and smoothness of the leaf blade led report. and petiole length. The largest ovaries with the most ovules occur in west central Amazonia; intermediate sizes and numbers are widespread MATERIAL AND METHODS but reach the coast only between Marajó and Ceará; and the smallest ovaries with the fewest The study was carried out with locules and ovules are coastal or nearcoastal from Delta Amacuro in Venezuela to Marajó. pressed specimens borrowed from many Small ovaries also occur in coastal Alagoas and herbaria, to whose curators I am grateful. at Rio de Janeiro. Ovaries with the fewest locu­ The most instructive characters are those les and ovules are believed to be the most of the unripened ovary, and therefoie specialized, the result of evolution toward only flowering material was of value in decreased waste of ovules, since the fruits of all members are few-seeded. Leaf characters most cases. It was necessary that speci­ correlate statistically with ovule numbers. mens have a considerable excess of flo­ Possible origen of the distribution pattern of wers for the dissections to be made wi­ the species is compared in terms of present thout harm but fortunately only a few rainfall patterns and in terms of Pleistocene climatic change with associated forest refuges. collections had to be omitted for this It is concluded that both phenomena were reason. In a few critical fruiting collec­ probably influential. High specialization appe­ tions leaf measurements alone were recor­ ars to have accompanied isolation, for reasons ded. The omission of most fruiting that are unclear. Because the plants from Delta collections from Fig. 10 does not materi­ Amacuro to Marajó are the most specialized they may once have been more isolated than ally affect the distribution shown, partly now. because most collections of the species are of flowering plants, partly because by INTRODUCTION good fortune most of the fruiting collec­ tions are from localities where flowering Mouriri guianensis is a small some­ material has also been taken or from times shrub-like tree of the neotropics. places nearby. No important range exten­ Its distribution lies across Brazil and sions are omitted. Coverage of the collec­ extends to Venezuela on the north and to tions I believe to be adequate for the Rio de Janeiro on the south (Fig. 10). purposes of this paper, but there are Certain characteristics of the plants are many gaps which will probably be filled highly variable over this range; the varia­ in the future. tion is examined here to see to what The relevant floral features are tho­ extent it may correlate with the geogra­ se of ovary size and shape (Fig. 3-9) and phic distribution. An unusual feature of the inner components that affect those the present example is that the direction characters. The diameter of the dried ( *) Botany Department, University of Minnesota, St. Paul, Minnesota, USA 55108. ovary can be used as one parameter; When the leaves are folded lengthwise no however, outside measurements of the reliable angle can be obtained. When all tapering and often distorted ovaries are are loose in a pocket an average angle can not always reliable. The ovary size is a of course be obtained but there is no reflection of the number of ovules within, certainty that the lowest and uppermost and to a lesser extent of the number and forms are present. Averages range from thickness of the partitions. Consequently 790 to 160O; the smallest and largest the numbers of locules and ovules have individual angles measured were 61° and been mapped as the major variables of the 180°. The basal angle is shown in Fig. 10 ovary. as a continuous variable. Material permitting, three to five Petiole lenght is also variable and flowers were dissected for each of the 71 has been mapped. The length ranges from collections examined and occasionally 1 to 5 mm, and the averages of the collec­ more. Limitations of material precluded tions vary from 1.5 to 4 mm. In this ins­ larger samples. In most cases the range of tance a two-state presentation seemed all variation encountered in a given collec­ that could be justified; a bar on the dia­ tion was not great with these small sam­ gram in Fig. 10 indicates a petiole average ple numbers, leading me to believe that greater than 2.5 mm. the results are reasonably representative. The prominence of the lateral The consistency of counts of different nerves in dried leaves is also greater in so­ collections from the same locality and the me areas than others and is mapped, consistency of the patterns shown in Fig. although sometimes highly variable and 10 support the opinion that the results difficult to measure consistently, only a are adequate for use. visual estimate being practical. The dried leaf surface was rated on a scale of 1—5, The leaves also vary with geography. from veiny to smooth, using a standard Blade shape is the most consistent varia­ veiny collection (Morley 1164) and a ble that is readily measured, and the smooth surface as reference points. In width of the basal angle is the most con­ this case also only a two-state presenta­ sistent character of the shape. As the an­ tion is given in Fig. 10. A bar on the gle becomes narrower, the leaves tend to diagram indicates a smoothness rating of become more elliptic and less ovate, but greater than 3. this does not always hold. The possibility that the leaves on a particular specimen The shoots of these plants are of do not represent a norm for that plant is two types, leader shoots and dorsiventral a problem that cannot be avoided. ones, and leaves of the two are dissimilar. The relatively few leaders, which grow Leaf variation on the same speci­ upward with great vigor, bear unusually men is often considerable. The leaf-bea­ short leaves which are often cordate at ring twigs grow in spurts or flushes, with base, while the much more common resting periods between. The lowest lea­ dorsiventral shoots produce the typical ves of each growth flush tend to have the leaves of the plant. Only the typical widest included basal angles, the upper leaves are referred to in this paper. the narrowest (Figs. 1„ 2). The difference may be very slight. When there is but a single pair of leaves on a twig, its leaves RESULTS are most often like those of the lowest leaves or sometimes the middle leaves of Examination of the distribution twigs with several pairs. The average angle map (Fig. 10) shows the following points is used here to express this leaf character. of significance: 1. Locule number. northeast tip of Marajó. C The coastal and near-coastal A. This varies from 5 (rarely 6) to collections in Alagoas and Rio one. Plants with 5 and 4 locules de Janeiro have relatively low are very^ widespread. locule numbers. The 2.6 that B. The smallest locule numbers occurs on the upper Rio Bran­ occur from coastal Delta Amacu- co is exceptional, by far the ro to the south side of Marajó. smallest number in the interior. All those ovaries with 2.5 or 2. Ovule number. fewer locules occur here, and none here have more than 3 ex­ A. This correlates with locule num­ cept for the collection at the ber, with some variation. Five 3 4 5 6789 Figs. 1, 2. Leafy twigs illustrating variation in shape of leaf blade from plant to plant and from lower to upper leaves in a growth flush. Tracings. 1. Moore 437. 2. Broadway 275. Figs. 3—9. Inferior ovary and calyx of different flowers illustrating variation in size and shape of the ovary in dried flowers. Camera luci- da. 3. Ule 5915. 4. Level 118. 5. Kubitzki 75-79. 6. Weddell 2494. 7. Froes 2008. 8. Oliveira 2082. 9. Soubirou s.n. locules are usually associated G.The only other occurrences of with 15 or more ovules, never 10's and 11's below 11.5 are less than 11. The smallest ovule also coastal or nearly so, one in numbers occur only in ovaries Alagoas and two at Rio de with 1 or 2 locules. Janeiro. The exceptional colle­ B. The greatest number of ovules ction from the upper Rio Branco occurs in west central Amazonia. with 11.6 ovules (referred to in The largest number counted point 1D) is the inland collection was 33, for the plant with closest in number to any of an average of 29.3. The these: five of its flowers yielded numbers decrease to the north, ovule counts from 10 to 13. east, and south from here. An 3. Ovary diameter: the fewer the area of moderately high numbers ovules and locules, the smaller occurs along the Amazon in Pará the ovary. Size differences are C. High-intermediate ovule num­ slight but rather consistent. Thus bers of about 20 occur in wes­ the largest ovaries are found in tern Venezuela, southwest Ama­ western Amazonia and the smal­ zonia, two places northwest of lest ones on or near the coast Brasilia, and at one site in inner from Georgetown to Amapá.