404 George E. Nichols, 



below they grow above, at the same time branching more or less 

 profusely. And as the branches become covered over they pro- 

 duce copious adventitious roots, with the result that the original 

 number of physiologically independent individuals, as viewed at 

 the surface of the hummock, becomes multiplied many times. 

 The shrubs, therefore, which cover the surface of a hummock 

 have been derived directly, in large part at least, from pre- 

 existing shrubs : they antedate the hummock itself." 



The genetic relationship between dwarf shrub and sedge- 

 grass heath, on the one hand, and dwarf shrub-spruce heath, 

 on the other, has already been suggested. During the evolution 

 of the present association-type, various of the herbaceous vascular 

 plants characteristic of the more primitive stages either disappear 

 or else become in large part or wholly confined to the depres- 

 sions : to situations where there is no great depth of humus and 

 where the soil relations presumably are more favorable than on 

 the hummocks. This is true, for example, of both Scirpiis and 

 Calamagrostis, and of such forms as Lycopodium, Potentilla, 

 Solidago, and Aster. At the same time, scrubby trees become 

 increasingly conspicuous. 



In its ecological aspect, an association of the sort just depicted 

 certainly approximates very closely dwarf shrub heath, as defined 

 by Warming ('09, pp. 210-214). It agrees in the nature of the 

 underlying soil, in the dominance of lichens and ericaceous 

 shrubs, in the low stature of the vegetation, and in the copious 

 production of raw humus : this latter a phenomenon which, 

 according to Warming, "must be regarded as the most charac- 

 teristic peculiarity of heath." It would seem to differ from typical 

 heath primarily in the presence of various arborescent species 

 which, in more favorable situations, attain much larger dimen- 

 sions than here. But while any of the trees of the climax 

 coniferous forest (with the exception of yellow birch and red 

 maple) may be represented here, it is significant that black spruce 

 is invariably predominant; that tamarack, which is practically 

 absent from the climax forest, is usually a prominent constituent ; 

 and that, on the other hand, balsam fir, the predominant tree 

 in forests of the regional climax type, is of very subordinate 



In this connection, see observations by Ganong, quoted on p. 447. 



