THE ECOLOGY OF TESTATE AMEBAS (PROTOZOA, RHIZOPODA) IN OCEANIC PEATLANDS IN NEWFOUNDLAND, CANADA - MODELING HYDROLOGICAL RELATIONSHIPS FOR PALEOENVIRONMENTAL RECONSTRUCTION
Dj. Charman et Bg. Warner, THE ECOLOGY OF TESTATE AMEBAS (PROTOZOA, RHIZOPODA) IN OCEANIC PEATLANDS IN NEWFOUNDLAND, CANADA - MODELING HYDROLOGICAL RELATIONSHIPS FOR PALEOENVIRONMENTAL RECONSTRUCTION, Ecoscience, 4(4), 1997, pp. 555-562
Peatland testate amoebae are important indicators of hydrology in foss
il studies. The direct quantitative application of this principle has
been limited due to a lack of suitable modern data and statistical tec
hniques. This study gives results on the ecological relationships of t
estate amoebae faunas from Newfoundland, Canada, and explores techniqu
es for modelling the relationship with water table depth. Water table,
soil moisture and pH are the dominant influences on species compositi
on in 60 samples from 14 peatland areas. Forty species occurred in > 6
samples and these are used to model the relationship with water table
depth. Several models were tested. Tolerance downweighted weighted av
eraging (WA-Tol), weighted averaging with partial least squares (WA-PL
S) and partial least squares (PLS) were assessed by jack-knifed error
estimates. WA-Tol out-performed all other models and had an RMSE of 6.
32 cm with a maximum bias along the gradient of 5.20 cm. WA-Tol probab
ly performs better because the relatively high tolerance values of som
e species have less influence on the reconstructed values. A series of
good indicator species can be identified on the basis of their narrow
tolerances. These are Arcella discoides, Difflugia bacillifera, Nebel
a carinata, Cryptodifflugia sacculus, Nebela griseola, Nebela marginat
a, Quadrulella symmetrica, Amphitrema stenostoma, and Sphenoderia lent
a. The analyses suggest that accurate quantitative reconstructions of
past water tables are possible using these techniques and that experim
entation with different models is worthwhile to improve their predicti
ve capacity. In order to avoid problems with poor analogues, reconstru
ctions should be based on larger more comprehensive data sets of moder
n faunas from a wider region.