Throughout the eastern United States, plant species distributions and commu
nity patterns have developed in response to heterogeneous environmental con
ditions and a wide range of historical factors, including complex histories
of natural and anthropogenic disturbance. Despite increased recognition of
the importance of disturbance in determining forest composition and struct
ure, few studies have assessed the relative influence of current environmen
t and historical factors on modern vegetation, in part because detailed kno
wledge of prior disturbance is often lacking. Ln the present study, we inve
stigate modern and historical factors that control vegetation patterns at H
arvard Forest in central Massachusetts, USA. Similar to the forested upland
s throughout the northeastern United States, the site is physiographically
heterogeneous and has a long and complex history of natural and anthropogen
ic disturbance. However, data on forest composition and disturbance history
collected over the past > 90 years allow us to evaluate the importance of
historical factors rigorously, which is rarely possible on other sites.
Soil analyses and historical sources document four categories of historical
land use on areas that are all forested today: cultivated fields, improved
pastures/mowings, unimproved pastures, and continuously forested woodlots.
Ordination and logistic regressions indicate that although species have re
sponded individualistically to a wide range of environmental and disturbanc
e factors, many species are influenced by three factors: soil drainage, lan
d use history, and C:N ratios. Few species vary in accordance with ionic gr
adients, damage from the 1938 hurricane, or a 1957 fire. Contrary to our ex
pectation that the effects of disturbance will diminish over time, historic
al land use predicts 1992 vegetation composition better than 1937 compositi
on, perhaps because historical woodlots have become increasingly differenti
ated from post-agricultural stands through the 20th century.
Interpretations of modem vegetation must consider the importance of histori
cal factors in addition to current environmental conditions. However, becau
se disturbances such as land use practices and wind damage are complex, it
is often difficult to detect disturbance effects using multivariate approac
hes, even when the broad history of disturbance is known.