DEFORMATION AND GRADATION. 311 



able places, a new though limited sere will renew the climax commmiity, which 

 will persist until destructive erosion again occurs. In most places, the new 

 sere will be destroyed in the initial or medial stages, and a second attempt at 

 development may follow. Similar denudation and succession may recur 

 several to many times, but the final outcome will be the stabiUzing of plain 

 and slope to a point where this oscillation is no longer possible, and the sere 

 runs its entire course to end in the cUmax. In the case of prairie, the general 

 process is the same, but the gradation of the slopes and the valley plain usually 

 results in the development into a scrub or woodland postclimax. 



The lateral planation wrought by a stream along its course does not differ 

 in essence from the tangential erosion typical of terminal and lateral gullies. 

 In both cases, erosion destroys vegetation, bares a slope or cliff, and is followed 

 by the deposition of the detritus to form a new area. In gullies, the deposit 

 is often made at the foot of the slope, and the area for invasion is small and 

 quickly reclaimed. Where banks are undercut by a stream, much or all of 

 the talus is removed and deposited in flats below. Even here there is a char- 

 acteristic association of erosion and deposit, since each concave bank due to 

 undercuttiag faces a convex flat bank formed by deposit. The erosion bank 

 behaves essentially like a guUy with respect to succession, particularly when 

 the imdercutting produces a steep or vertical cliff. The formation of a talus 

 often deflects the current from such a bank, and when it returns it xmdercuts 

 a slope covered with vegetation. Thus the erosion edge may be at the water- 

 level, and then is the same as the cutting-edge, but usually it is determined by 

 the cliff-edge produced by the cutting of the stream. As long as a vertical 

 cliff is maintained by cutting, only a slight initial development is possible, and 

 this is mostly of Uchens and mosses on rock-faces. As a rule, however, such 

 cliffs are broken by gullies and pockets, in which the sere may reach partial or 

 complete development, and from which it may spread slowly into other areas. 

 If undercutting stops for a sufficient time, the development reestablishes the 

 climax. When it begins again the vegetation, whether climax or serai, is 

 destroyed, and a new area for invasion is produced. Thus the stream-bank, 

 like the guUy, may exhibit a single sere, or, more regularly, it shows a frag- 

 mented cosere which terminates in the climax. The planation of flood-waters 

 and their effect upon vegetation are practically the same as that of the normal 

 stream, though the action is intermittent rather than continuous. 



Since the erosion bank and the deposit flat are separated by the stream, and 

 since the initial conditions are diametrically opposite, their relation to succes- 

 sion is very different. While both must eventually terminate in the same 

 climax, the one is hydrarch, the other xerarch. The sere on the deposit bank 

 will pass regularly into the climax condition, which will persist until a swing 

 of the current begins to remove it by undercutting, or until it is destroyed by 

 flooding. The latter may of course occur repeatedly, and result in a series 

 of incomplete seres, i. e., a fragmented cosere. Just as flood-waters may erode 

 bluffs and terraces, so they may also fill portions of the flood-plain, destroying 

 vegetation and producing new areas for invasion (plate 56, a, b). 



The relation of a local or regional cycle of erosion to succession must have 

 been the same in the past as it is at present. Its general effect must have 

 been to destroy the climax gradually and to produce new areas for its repro- 

 duction. While areas of excessive erosion, such as bad lands, must have 



