Diatoms as indicators of historical macrophyte biomass in Florida lakes

MISSING IMAGE

Material Information

Title:
Diatoms as indicators of historical macrophyte biomass in Florida lakes
Physical Description:
vii, 136 leaves : ill. ; 28 cm.
Language:
English
Creator:
Whitmore, Thomas J ( Thomas James ), 1955-
Publication Date:

Subjects

Subjects / Keywords:
Diatoms -- Florida   ( lcsh )
Zoology thesis Ph. D
Dissertations, Academic -- Zoology -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1991.
Bibliography:
Includes bibliographical references (leaves 123-134).
Statement of Responsibility:
by Thomas James Whitmore.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001683410
oclc - 25034703
notis - AHZ5381
System ID:
AA00004755:00001

Full Text










DIATOMS AS INDICATORS OF HISTORICAL
MACROPHYTE BIOMASS IN FLORIDA LAKES








By



THOMAS JAMES WHITMORE


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1991













ACKNOWLEDGMENTS


I thank Daniel E. Canfield, Jr. for suggesting the concept of this

study in constructive criticisms he offered during my master's
defense. Dan's challenge provided me with motivation: I chose this
topic because I was determined to show him I could resolve the
problem. Mark Hoyer and Christine Horsburgh were responsible for
collection and management of macrophyte data, and Mark was
especially helpful in providing me with this information. The Florida
Museum of Natural History provided the vehicles, boats, and
equipment for field work. Brian Rood, J. Daniel Bryant and Dennis
Crumby assisted in the field. Silvia Ferguson gave much time and
hard work in field assistance, and has been a constant source of
support throughout this study.
I am grateful to Claire Schelske, Marty Fleisher, and Arthur

Peplow for radionuclide assay of samples used to calculate bulk

sediment accumulation rates. I thank Douglas Turley for the time
and computer expertise he offered for data management. Paul Zimba
gave generous assistance in computer programming during his
personal time, making it possible for me to use the CANOCO statistical
package. Dana Griffin provided funds for mainframe computing.
Martha Love and Sarah E. Whitmore kindly helped with manuscript
preparation.









This dissertation work was begun under the guidance of Edward

S. Deevey, Jr., my major professor for more than 9 years. Ed's death
in November 1988 was a blow to many academic disciplines, and a
sad personal loss of which I am especially cognizant at this time. I
have benefitted from Frank Nordlie's particulalry competent
guidance through the completion of my studies. Claire Schelske
generously provided the space, resources and financial support that
allowed me to complete my graduate work. Joseph S. Davis and
Ronald G. Wolff have been supportive members of my graduate
committee for many years. It is largely due to Ron's encouragement
that I pursued graduate studies at the University of Florida. No list
of acknowledgements for my graduate work would be complete
without expressing sincere thanks for all manner of support and
encouragement to Mark Brenner, my colleague of 11-1/2 years who
introduced me to paleolimnology.













TABLE OF CONTENTS


Page
ACKNOW LEDGM ENTS........................................................................................... ii

A B STRA CT ........................................................................................... ................... vi

CHAPTER
1 INTRODU CTION ..................................................................................... 1

The Concept of Trophic State in Lakes.......................... ....... 1
Macrophytes and the Lake Ecosystem................................ 4
The Relationship of Macrophytes With Epiphyton............. 11
Methods for Reconstructing Historical Macrophyte
Communities................................. .................. 15
Diatom Methods in Paleolimnology.......................... ............ 16
Effects of Spatial Variation on Diatom Assemblages.......... 27
Purpose............................................ ............................................ .... 2 9

2 METHODS....................................................... 32

Collection of Sediment Samples............................................ 32
Laboratory Analyses.................................................... 33
Quantifying Macrophyte Presence............................. .......... 36
W ater Chemistry Data.................................................................... 38
Statistical Analyses......................................... ........................... 38

3 RESU LTS.............................................................. ........................... 4 3

Results of Cluster Analyses................................................... 6 1
Results of Principal Components Analyses.......................... 68
Results of Stepwise Multiple Regression.............................. 76
Results of Canonical Correspondence Analyses.................. 83
A New Predictive Model for Water-Column Total P.......... 85
Assessing Confoundedness in the Water-Column Total
P Predictive Model........................................ ............. 88











4 DISCUSSION ......................................................................................... 90

Dominant Environmental Variables and Scale
of A nalysis............................................................................ 9 0
Response of Periphyton to Water-Column Nutrients....... 93
Recommended Predictive Models for Macrophyte
Variables..................................................................... 95
Applying Predictive Models to Obtain Historical TSI
Estim ates..................................................................... 98

APPENDIX 1.......................................... ........... 101

APPENDIX 2........................................................ 102

APPENDIX 3................................................... 109

APPENDIX 4.................................................................................. 116

A PPEN D IX 5 ........................................................................................................ 1 1 8

APPENDIX 6............................................ ..... ..... 120

BIBLIOGRAPHY......................................................................................... 123

BIOGRAPHICAL SKETCH................................. .............. 13 5













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

DIATOMS AS INDICATORS OF HISTORICAL
MACROPHYTE BIOMASS IN FLORIDA LAKES
By
Thomas James Whitmore
May 1991

Chairman: Frank G. Nordlie
Major Department : Zoology


Macrophytes represent an important component of primary
production in lakes that is usually ignored in trophic state
classification. Trophic classifications traditionally emphasize water-
column nutrient concentrations and phytoplankton biomass.
Predictive models have been developed from diatom assemblages to
assess historical changes in lake trophic state, but these models
usually infer water-column total P or chlorophyll a values and thus
also ignore macrophyte production. Lake-sediment core samples
often indicate former periods of low trophic state, although these
periods may instead represent low water-level events that
periodically occur in Florida lakes. Because macrophyte biomass is
negatively correlated with water-column nutrients, macrophyte
biomass may have been high at times when nutrient inferences
suggest that lakes were unproductive.








The purpose of this study was to develop predictive models for
inferring historic macrophyte biomass using diatoms, and to
incorporate those models into a scheme that permits a more
complete assessment of former lake trophic state than do models
based solely on water-column nutrient concentrations.
Subfossil diatom assemblages were analyzed from the surface
sediments of 29 Florida lakes covering a range of macrophyte
abundance. Trophic-state, pH, and specific conductance formed an
environmental gradient that was the principal influence on diatom
communities in the limnologically diverse set of lakes. The
planktonic proportion of the diatom community was positively
correlated with trophic state, whereas the periphytic proportion was
negatively correlated with trophic state. Sedimentary diatom
concentrations, however, showed that both of these life-form
communities had a positive response to increase in water-column
nutrients.
Multivariate models are presented that permit estimates of
former macrophyte biomass from fossil diatom assemblages. The kg
of P contained in once-living macrophyte biomass can be estimated
using a mean percent P value for macrophyte taxa. This mass of P
divided by lake volume yields a concentration that can be added to
limnetic P inferences obtained from diatom predictive models to
estimate the potential total P content of the water column (WCP)
Trophic state index values calculated with historic WCP will reflect
both former macrophyte and phytoplankton aspects of trophic state.













CHAPTER 1


INTRODUCTION


The Concept of Trophic State in Lakes


Primary productivity in lakes can be defined as the rate at
which new organic matter is formed by photosynthesis in autotrophs
such as phytoplankton and macrophytes. Annual productivity is
typically expressed as the number of grams of carbon fixed per unit
of lake surface area per year. Lake productivity, however, has been
more traditionally thought of in conceptual terms referred to as
trophic state, and typologically described using categories ranging
from ultraoligotrophic at the low end of productivity to
hypereutrophic at the high end (cf. Shannon and Brezonik 1972).
Several trophic state indices (TSI) have been developed that
permit numerical expression of trophic state using biological,
physical and chemical characteristics. Shannon and Brezonik (1972),
for instance, used principal components analysis to reduce seven

variables including chlorophyll a (Chl a), primary productivity, and
total P to a single variable that described the trophic status of lakes.
Although it was inclusive, this TSI has been regarded as
cumbersome, especially because of the difficulty in obtaining
primary productivity values. Carlson (1977) sought a single, easily
obtained measure to describe lake trophic state, and he selected








Secchi depth, a convenient variable that reflected phytoplankton
standing crop in many lakes. Carlson scaled his TSI with the
intention that a ten-unit increase in TSI would be equivalent to a
doubling of algal biomass, but his index failed to relate in a uniform
way to Chl a. Carlson calibrated his TSI to limnetic Chl a and total P
for a set of north-temperate lakes and developed subindices to
permit expression of TSI from these latter two variables.
Kratzer and Brezonik (1981) later modified Carlson's approach
by constructing a subindex expressing TSI from water-column total N
concentrations after observing N limitation in several Florida lakes
located on phosphatic limestones. They proposed an averaged
subindex that used mean TSI values based on Chl a and Secchi depth,
and the lesser of the two TSI values based on total P and total N.
Baker et al. (1981) observed different relationships between
Secchi depth, total P and Chl a in Florida lakes than Carlson (1977)
observed in north-temperate lakes. Huber et al. (1982), therefore,
constructed a new TSI for Florida lakes using the Florida Lakes Data
Base, a large data set that is maintained at the Water Resources
Research Center at the University of Florida.
Huber et al. based their TSI on Chl a, a more direct measure of
phytoplankton biomass than Secchi depth, and they retained Kratzer
and Brezonik's total N and averaged subindex approach because of N
limitation in some Florida lakes. Huber et al. described lakes with
total N/total P values >30 as P-limited, and calculated averaged TSI
(TSI(AVG) for these lakes as the mean of TSIs based on Secchi depth
(TSI(SD)), Chl a (TSI(Chl a) and total P (TSI(TP)). Lakes with total
N/total P values <10 were described as N-limited, and TSI(AVG) was








calculated as the mean of TSI(SD), TSI(Chl a) and TSI(TN). Huber et
al. regarded lakes with total N/total P values between 10 and 30 as
nutrient-balanced, and noted that nutrient-Chl a relationships were
different in these lakes than in nutrient-limited lakes. They
constructed new TSI expressions for total P (TSI(TPB) and total N
(TSI(TNB) for nutrient-balanced lakes, and defined TSI(AVG) for
these lakes as the mean of TSI(SD), TSI(Chl a), TSI(TPB) and
TSI(TNB). The TSI of Huber et al. is the only TSI developed
specifically for use in Florida lakes.
All of the trophic state indices discussed above, however, share
in common their bias towards phytoplankton biomass and water-
column nutrient concentrations as the relevant indicators of primary
production in lakes, and they give no importance to the presence of
macrophytes. Porcella et al. (1980) constructed a multivariate index
based on Carlson's (1977) subindices and included a term derived
from percent-area coverage of macrophytes. Because Porcella et al.'s
TSI was developed for north-temperate lakes that are P-limited and
demonstrate hypolimnetic oxygen deficits during stratification, this
TSI might be inappropriate for use in Florida. The macrophyte term
also failed to quantify nutrients contained in macrophyte biomass as
other TSIs using water-column total P quantify the nutrients
contained in phytoplankton biomass.
Canfield et al. (1983a) proposed a new approach to trophic state
classification of lakes that gave consideration to nutrients contained
in macrophyte biomass as well as to water-column nutrient
concentrations. They quantified the amount of P contained per unit
of dry weight in many species of macrophytes from 6 Florida lakes.









Next they estimated the macrophyte biomass in each lake from the
area covered by macrophytes and the macrophyte density in kg of
wet biomass per square meter. This permitted calculation of the kg
of P contained in macrophyte biomass in each lake and the amount of
P that would be released to the water column assuming 100% death
and decomposition of the macrophytes. Dividing this mass of P by
the lake volume yielded a concentration that when added to water
column total P produced an estimate of the potential total P content
of the water column (WCP). This approach provided more realistic
estimates of trophic state for lakes such as Fairview in Orange Co.,
which appeared oligotrophic based on water-column nutrients and
Chl a, but contained a large standing crop of macrophytes. The
estimate of total water-column P content brought this lake to the
eutrophic range, which was edaphically consistent with other lakes in
the same physiographic region. Recent evidence shows that water-
column P concentration in Lake Fairview has increased to the
predicted WCP level because of macrophyte removal by grass carp
(Canfield pers. comm.). Canfield et al. (1983a) noted that the same
approach may be used with N for lakes that are N-limited.


Macrophytes and the Lake Ecosystem
Despite the emphasis on water-column nutrients and
phytoplankton biomass to characterize lake trophic state,
macrophytes are responsible for a substantial amount of the primary
production that occurs in many lakes. This is especially the case in
Florida because the shallow depths of Florida lakes, the high amounts
of insolation and the long growing season are conditions that support








high macrophyte standing crops (Brenner et al. 1990). Many Florida
lakes have high nutrient concentrations because of edaphic reasons
or anthropogenic loading (Canfield and Hoyer 1988), and this also
stimulates macrophyte production.


Macrophyte Growth Forms
Macrophyte species are often grouped into growth-form
categories that describe whether or not the plants are rooted in
sediments and whether they grow laterally in the water or erect and
out of the water. Submerged macrophytes are those typically rooted
in sediments, growing completely under the water and usually
flexible due to a lack of rigid cellular tissue. Myriophyllum
heterophyllum Michx., Utricularia purpurea Walt. and Ceratophyllum
demersum L. are three examples of submerged taxa native to Florida,
while another common taxon, Hydrilla verticillata Royle is an
introduced exotic that has proliferated widely. Many submerged
taxa, when growing in dense stands, are regarded as nuisance species
that have a negative effect on lake recreational uses (Brenner et al.
1990).
Floating-leaved plants can be divided into two categories
depending on whether they are rooted in sediments or not. Rooted
floating-leaved plants derive most of their nutrients from the
sediments (Carignan and Kalff 1980) and often have large peltate
leaves growing at the surface where they have access to sunlight and
atmospheric C02 for photosynthesis. Common examples of these taxa
found in Florida are Nymphaea spp. (water-lily), Nelumbo lutea
(Willd.) Pers. (American lotus), Nuphar luteum (L.) Sibth. & Smith








(spadderdock), Nymphoides aquatica (Gmel.) O.Ktze and Brasenia
shreberi Gmelin. A second group of floating-leaved taxa are
unrooted in sediments and free-floating. These taxa, which include
Lemna minor L. (duckweed), Pistia stratiotes L. (water-lettuce) and
Salvinia rotundifolia Willd., obtain their nutrients from the water and
exhibit adaptations that keep the plant afloat. Eichhornia crassipes
(Mart.) Solms. is a floating-leaved species introduced to Florida,
which because of its rapid spread and prolific growth, has become a
severe economic and environmental problem (Tarver et al. 1979).
Emergent taxa, which grow erect in shallow aquatic areas and do
not depend on the water for support, demonstrate the third growth
form in macrophytes. Common examples of these taxa are Typha
spp. (cattails), which was the macrophyte with the most extensive
areal coverage in a large survey of Florida lakes (Schardt 1983), and
Sagittaria latifolia Willd.
Some taxa exhibit growth patterns that are typical of more than
one growth-form category. Hydrocotyl umbellata L., for instance,
grows mostly as a submerged plant though leaves are frequently
emergent in shallow water. Portions of Hydrocotyl mats occasionally
break away and are redistributed as floating vegetation.
Potamogeton spp. also exhibits extensive lateral submerged growth,
but bears some floating leaves at the surface.


Environmental Factors Influencing Macrophyte Distribution
Many studies have been conducted to determine which
environmental factors most affect the distribution and abundance of
macrophytes, and many of these studies have come to different








conclusions. Collins et al. (1987), for instance, compared macrophyte
biomass density with 13 different chemical, physical and biological
variables at various sites in Lake George, NY. They concluded that
water depth was the most important factor affecting macrophyte
biomass, and that substrate type and eutrophication status were of
secondary importance. Canfield and Hoyer (1988) studied the
influence of light and nutrient availability on macrophytes in Florida
streams, and they concluded that nutrients do not regulate the
abundance of macrophytes. Shading of macrophytes was the most
important factor regulating macrophytes in that study, while
substrate type, water depth and current velocity had a secondary
influence. Jackson and Charles (1988) studied macrophyte species
composition in 31 small, unproductive lakes in New York that were
low in specific conductance. They concluded that pH was the
regulating factor, that area, slope and substrate composition were of
secondary importance, and that macrophyte distribution bore no
relation to trophic state indicators. Crowder et al. (1977) concluded
that specific conductance was as important a macrophyte
determinant as pH in their study on circumneutral to hardwater
lakes. Duarte and Kalff (1990) determined that alkalinity and slope
were the most important factors in their study, but they explained
the discrepant conclusions between studies as the result of
differences in scale of analysis. When macrophytes are compared
between lakes in hardwater areas, water chemistry, including
specific conductance and trophic state are bound to be important
determinants of macrophyte distribution (Duarte and Kalff 1990,
Jackson and Charles 1988). Surveys of unproductive, dilute lakes








cover a small range of difference along the pH-alkalinity-
conductivity complex, and they are likely to conclude that pH is the
important variable affecting macrophyte distribution (Jackson and
Charles 1988). Surveys conducted within one or a few lakes will
cover only a small range of water chemisty differences, and site
characteristics, such as waves, slope and sediment type, will prove to
be the most important determinants (Duarte and Kalff 1990). Within
a single lake, wave exposure is likely to be a leading determinant of
macrophyte biomass at shallow littoral depths, whereas water
transparency will exert more influence at greater depths (Duarte and
Kalff 1990).


Effect of Macrophytes on Lake Ecosystem
Macrophytes seem to exert considerable effects on the nutrient
cycling, biology, sedimentation patterns and senescence of the lakes
in which they occur. Rooted macrophytes obtain most of their
nutrients from lake sediments and thus link the sediments with
overlying water (Carpenter 1981). This provides a mechanism for the
regeneration of sedimentary nutrients into the water-column.
Carignan and Kalff (1982) observed that living macrophytes were
responsible for a 2.2% daily increase in P that represented a net
seasonal input to the littoral zone because the P was derived from
sediments. While P is not released from living macrophytes at a
rapid rate, substantial amounts of nutrients in the macrophyte
biomass are released when macrophytes die back and shoots decay.
Approximately 75% of the P released is in a soluble reactive form,
and P becomes rapidly assimilated by phytoplankton, leading to an








increase in water-column Chl a (Carpenter and Lodge 1986). Landers
(1982) estimated that approximately 18% of the annual P loading in
Lake Monroe, Indiana originated from senescing macrophytes.
Carpenter (1981) also concluded that most of the dissolved organic
carbon and dissolved total P in Lake Wingra, Wisconsin was released
during decomposition of Myriophyllum spicatum in the littoral zone.
Filbin and Barko (1985) have concluded that the release of
sedimentary nutrients into the water column by macrophytes may
be more significant in lakes than in reservoirs because of the
riverine nature of reservoirs.
Macrophytes have several influences on the sedimentation
patterns in lakes where they are found. Macrophytes tend to
intercept or modify the flow of materials such as sediment from land
to the pelagic zone. By reducing water velocity and wave action,
macrophytes function as sediment traps in the littoral zone. This
effect was shown to be significant in historical changes in
sedimentation patterns of Lough Augher, Northern Ireland
(Anderson 1990b). When macrophytes die, their biomass increases
sedimentary organic matter content and leads to an accretion of
littoral sediment that promotes infilling of the lake basin and
expansion of emergent vegetation (Carpenter 1981, Carpenter and
Lodge 1986). Macrophyte presence in lakes, therefore, accelerates
infilling and senescence of lakes.
Macrophytes provide a complex habitat, and their presence leads
to an increase in those species commonly found in littoral areas.
When macrophytes are present, epiphytic algae proliferate and an
increase is observed in epiphytic grazers such as snails (Carpenter









and Lodge 1986). Zooplankton are abundant in weed beds and the
habitat complexity also provides cover and protection for spawning
and young fish (Carpenter and Lodge 1986). Dense infestations of
submerged macrophytes, nevertheless, have been shown to have a
negative effect on the presence of sport fish (Shireman and Maceina
1981).
Conflicting reports have been presented about the effects of
eliminating macrophytes in lakes through chemical or biological
control. Carpenter and Lodge (1986) stated that because of the
macrophyte role in enhancing sedimentary P recycling, an increase in
macrophyte standing crop will lead to an increase in phytoplankton
standing crop, whereas the long-term effect (>3 yrs.) of killing
macrophytes will lead to a decrease in water-column N and P and a
decrease in phytoplankton. This positive correlation between
macrophyte and phytoplankton standing crop is contrary to the
negative relationship reported by Canfield et al. (1984) between the
percent of lake volume infested with macrophytes and water-column
Chl a for 32 Florida lakes. An increase in water-column P
concentrations and phytoplankton standing crop has been shown
following herbicide application to macrophytes in Florida lakes
because of nutrient release by the decaying plant material (Richard
et al. 1984). An increase in water-column P was also reported
following biological control of macrophytes using the grass carp
Ctenopharyngodon idella because of nutrient release from feces,
although this increase seems less dramatic because of the retention
of P in the fish biomass (Richard et al. 1984, Canfield et al. 1983b,
Canfield et al. 1984).










The Relationship of Macrophytes With Epiphyton
Epiphytic algae growing in macrophyte beds often exhibit high
concentrations of biomass and are responsible for a significant
proportion of the primary production in a lake. Allen and Oceuski
(1981), for instance, determined that algal epiphytic production in
Lake Ohrid, Yugoslavia was higher than the production they
observed in littoral or pelagic algae. Cattaneo and Kalff (1980)
observed that the epiphytic algae in eutrophic portions of Lake
Memphremagog, Quebec fixed more carbon than macrophytes did
throughout the growing season. Fontaine and Ewel (1981) estimated
that macrophytes and their associated epiphytes were responsible
for 56% of the gross production in Little Lake Conway, Florida.
The question of macrophytes as a nutrient source for their
epiphytic algae has been a much-debated issue often referred to as
the "neutral substrate hypothesis" in the literature. Cattaneo and
Kalff (1979) observed no significant difference in epiphytic
production on Potamogeton richardsonii and artificial plants made of
plastic. They concluded that macrophytes functioned as neutral
support structures. Carignan and Kalff (1982) studied epiphytic
algae growing on fully 32p labelled Myriophyllum spicatum and
concluded that epiphytes derived only 3.4-9.0% of their P from the
labelled macrophytes, and that macrophytes were more important to
epiphytes for support than as a P source. Gough and Gough (1981)
took issue with Cattaneo and Kalff (1979), and cited Hutchinson's
(1975) statements that macrophytes release by-products of nutrient
assimilation, photosynthates and inorganic nutrients. They argued








that although some macrophytes may be neutral hosts, others affect

epiphytic production or community composition. Cattaneo and Kalff
(1981) replied that epiphytic biomass was mostly related to the
surface area of the substrate on which the epiphytes grow, and that
water chemistry exhibits a greater influence than macrophytes on
epiphytic production.
Recent studies by Burkholder and Wetzel (1990) seem to offer a
more definitive explanation of macrophyte influence on epiphytes.
They measured alkaline phosphatase (APA), an enzyme that
catalyzes hydrolysis of organic P compounds to release
orthophosphate, in epiphyton growing on natural and artificial
plants. They observed, as did Cattaneo and Kalff (1979), that APA
concentrations were higher in epiphyton growing on artificial
substrates than they were in epiphyton growing on macrophytes.
Burkholder and Wetzel concluded that epiphyton on artificial
substrates are P-limited and synthesize APA to provide a P source,
although the epiphyton growing on macrophytes were not P-limited
because of nutrient release by the macrophytes.
Epiphyton can in turn exert effects that influence the growth of
their macrophyte hosts. Nutrients are usually in abundant supply to
rooted macrophytes, and macrophytes generally do not seem to
compete with epiphyton for this resource. When epiphyton biomass
is high, however, epiphyton may shade their macrophyte hosts
(Eminson and Moss 1980). Filbin and Barko (1985) observed that

epiphytic biomass in Eau Galle Reservoir, Wisconsin comprised up to
33% of the macrophyte and epiphyte biomass, and they concluded
that epiphyton may have limited macrophyte growth by light









attenuation. Sand-Jenson and Sondergaard (1981) studied
phytoplankton and epiphyton shading effects on macrophytes in
Danish lakes. In oligotrophic, silicate-poor lakes, the water was
responsible for most of the light attenuation. Epiphyton were
responsible for 50% of the light attenuation to macrophytes in
oligotrophic, silicate-rich lakes receiving N supply. In a lake that had
a high nutrient supply, they determined that epiphytes were
responsible for 86% of the light attenuation to macrophytes.
Sand-Jensen and Sondergaard concluded that the shading effects that
epiphytes exert on macrophytes becomes a decisive factor limiting
depth distribution of macrophytes in lakes with high nutrient supply.


Substrate Specificity and Growth Forms of Periphyton
Some studies have indicated a high degree of substrate
specificity by epiphytic and periphytic diatoms. Round (1956)
characterized diatom taxa growing on plants (epiphytic) as
"attachment" types mostly of the genera Achnanthes, Cymbella and
Epithemia, whereas diatoms found on sediments (epipelic) were
actively motile and unattached, including the genera Navicula,
Amphora and Diploneis. Round noted, however, that diatom taxa
growing on stones (epilithic) were similar to epiphytic diatoms. Siver
(1978) observed that the diatom genera Achnanthes, Cocconeis and
Eunotia were the most abundant taxa growing on Potamogeton
robinsii. Blindow (1987) stated that the composition of epiphyton on
Potamogeton and Chara was different than the epiphytic composition
on Nitellopsis that was heavily marl-encrusted. Eminson and Moss
(1980) observed that host specificity of periphyton was greater in









oligotrophic lakes because of the importance of macrophyte nutrient
loss to epiphytes, whereas host specificity was less pronounced in
mesotrophic and eutrophic lakes because of the greater effect of
water-column nutrients on periphytic taxa.
Several studies have demonstrated the importance of diatom
growth form to patterns of colonization and physical structure of
epiphytic diatom communities. Achnanthes and Cocconeis are
solitary cells that lie adnate to the substrate and colonize
horizontally, and these genera are usually the initial colonizers on
new substrate (Robinson and Rushforth 1987). Later colonizers must
contend with space limitations, and genera such as Gomphonema and
Cymbella are at an advantage because they grow on long stalks and
colonize in a vertical orientation (Roemer et al. 1984, Robinson and
Rushforth 1987). This upward expansion of the epiphytic community
improves light and nutrient availability for taxa in the higher tiers
(Hudson and Legendre 1987), though some adnate forms below such
as Cocconeis placentula var. euglypta (Ehr.) Cl. exhibit shade
tolerance (Robinson and Rushforth 1987). Swift-moving taxa capable
of complex movements including Nitzschia and Navicula can be
observed within the community matrix (Hudson and Legendre 1987).
As the thickness of periphyton on the substrate becomes too great,
cells on the outer tiers are subject to loss by grazing or sloughing off
by currents (Roemer et al. 1984, Hudson and Legendre 1987).
Sloughed off periphytic taxa can become part of the planktonic drift
and are then referred to as tychoplanktonic (Lowe 1974).










Methods for Reconstructing Historical Macrophyte Communities
Historical macrophyte presence has been traditionally
determined from lake sediments by methods that do not yield
quantitative estimates of standing crop. Macrophyte presence has
been assessed historically from macrophyte remains, pollen and
seeds that are found in lake sediment. Davis (1985) studied
historical macrophyte presence in upper Chesapeake Bay and
summarized many of the biological and diagenetic factors that
obscure accurate reconstruction of former macrophyte communities.
Seed preservation is poor in some taxa (e.g. Vallisneria and
Potamogeton) and seed dispersal is poor in others (e.g.
Myriophyllum) leading to under-representation of these taxa in
sediments. Pollen and seed production is variable among species of
macrophytes (Yeo 1966, Birks 1980), and plants producing larger
quantities of these may be over-represented in the sedimentary
record. Seeds and pollen also may be unreliable indicators of
macrophyte standing crop because a large number of species
reproduce vegetatively by budding, fragmentation and by plants
arising from stolons and rhizomes (Tarver et al. 1979).
Seed representation in the sedimentary record may be affected
by differential transport and palatability (Birks 1980). Birks (1973)
and Watts (1978) have shown that seed dispersal is often localized
for macrophyte taxa. Dispersal patterns, therefore, can cause a high
degree of spatial variability of macrophyte indicators in lake
sediment. Sampling from many littoral sediment cores is required to
obtain a reliable reconstruction of macrophyte history.








To summarize, traditional methods of macrophyte community
reconstruction tend to over-represent, under-represent or miss
entire portions of the macrophyte community. No single method has
been developed that will provide reasonably accurate quantitative
estimates of historical macrophyte standing crop.


Diatom Methods in Paleolimnology
Some Quantitative Diatom Methods Used in pH Reconstructions
Most quantitative work using diatoms to reconstruct past
limnological conditions has been concerned with lake acidification
due to anthropogenically induced acid precipitation. The large
number of lake acidification studies recently funded (Davis 1987)
indicates that lake acidification has occupied an important place on
the agenda of national and international environmental concerns.
The high costs of implementing more rigid air pollution standards
necessitated statistical rigor to determine if atmospheric loadings of
sulfur and N oxides were having significant fallout effects on aquatic
ecosystems. As a consequence, lake acidification studies received
priority funding and were numerous. Davis (1987) reviewed many
such studies that used diatoms to infer historical pH trends.
The earliest quantitative index relating diatom assemblages to
pH of lakewater was the a index described by Nygaard (1956), a

ratio of acidic to alkaline diatoms in a sample based on the pH
autecological classifications (Hustedt 1937-38) of the individual taxa.
Renberg and Hellberg (1982) developed the somewhat more
sophisticated index B that was also a ratio of the percentage of
diatoms in pH autecological categories. These authors regressed log-








transformed index B values with pH for a set of 30 Swedish lakes
and produced a model with which they assessed lake acidification
due to atmospheric deposition in Sweden. Index B is somewhat
statistically dubious, however, because coefficients for the
autecological terms could not have been calculated by a simple linear
regression between pH and log index B as indicated by Renberg and
Hellberg (Whitmore 1989).
Cluster analysis has been used to identify diatom assemblages
characteristic of various pH conditions (e.g. Davis and Anderson
1985). Charles (1985) used cluster analysis to group diatom species
with similar pH requirements and he performed a multiple
regression of these clusters with pH values of 38 Adirondack lakes.
His model explained approximately 90% of the variance in pH in his
calibration data set.
Davis and Berge (1980) performed a stepwise multiple
regression of 33 taxa in a set of Norwegian lakes and produced a
model consisting of 7 taxa that explained 93% of the variance in pH
(unadjusted R2). Dixit and Evans (1986) have shown, however, that
particularly in lakes with spatial variability in diatom assemblages,
the use of indicator assemblages rather than individual taxa in
predictive models will greatly reduce the error term. Hustedt's
(1937-38) pH autecological categories have also been used in
multiple regression equations to develop pH predictive models based
on diatom assemblages (Davis and Berge 1980, Charles 1984, 1985).
Ordination techniques, which reduce the number of diatom
variables in a model, have been used to construct pH predictive
equations. Principal components analysis is an indirect ordination








technique that was used to develop models for assessing lake
acidification in Maine and Norway (Davis and Berge 1980, Davis and
Anderson 1985). Davis and Anderson (1985) found that principal
component models were less sensitive than multiple regression
models using individual taxa to variation in the frequencies of taxa
caused by environmental factors other than pH. Van dam et al.
(1980) also used a principal components procedure to calibrate
diatom models and assess the effects of acid precipitation on Dutch
moorland pools.
Reciprocal averaging (RA) is another indirect ordination
technique that has been used in diatom-based models. Charles
(1985) used reciprocal averaging to ordinate diatom data and he
correlated RA axes with environmental variables in a set of
Adirondack lakes. pH was a primary determinant of diatom
assemblage composition in that set of lakes and a regression equation
predicting pH from Charles' first RA axis explained 90% of the
variance in pH. In other environmental applications, Servant-
Vildary and Roux (1990) have used reciprocal averaging to
determine the effects of ionic elements on diatom species
composition in saline lakes of the Bolivian Altiplano.
Canonical correspondence analysis (CANOCO) is a direct
ordination technique that has been used in pH reconstructions to
define axes that are combinations of taxa responding directly to pH.
The axes have then been regressed with pH and the resulting models
used to document historical trends in lake acidification. Battarbee et
al. (1988), for example, used CANOCO to assess the acidification of








Scottish lochs and their recovery following abatement of atmospheric
sulfate emissions in the United Kingdom.


Diatom Methods for Reconstructing Trophic State
Historical trophic state studies generally have not received the
degree of quantitative treatment that studies of lake acidification
have. Diatom/trophic reconstructions have often relied heavily on
autecological information of specific taxa for qualitative
interpretation of diatom percentage diagrams from lake sediment
cores. Brugam (1978), for instance, documented the eutrophication
of Linsley Pond in Connecticut and Bradbury (1975) used diatoms to
interpret the history and eutrophication of Minnesota lakes.
Battarbee (1978) observed the influence of land use and sewage
effluent on the eutrophication of Lough Neagh, Northern Ireland.
HAkansson (1982) presented an excellent ecological analysis of the
diatom flora from HAvgArdssj6n in Sweden and documented
eutrophication after 1900 due to agricultural activity. Qualitative
studies have provided understanding of gross trends in the trophic
trajectory of lakes because of climatic patterns and anthropogenic
influence, but they have lacked ability to discern subtle trophic
differences, assess rates of change or demonstrate statistical
significance.
Ratios have been proposed that quantitatively describe lake
productivity using the percentages of diatom species separated at
high taxonomic levels. Nygaard's (1949) C/P index was a ratio of the
number of valves in the diatom orders Centrales and Pennales. High
C/P values were thought to indicate eutrophic conditions because of








the supposed eutrophic preference of Centrales, a notion refuted by
the wide range of trophic preferences actually observed for centric
taxa (Battarbee 1979). Stockner and Benson (1967) studied historic
trends in Lake Washington and proposed the A:C index, a ratio of the
number of valves in the tribe Araphidiniae to the number of valves
in the order Centrales. Centrales were assumed to be oligotrophic
rather than eutrophic indicators in this scheme. Stockner (1971)
later qualified the conditions under which this index would
accurately indicate trophic state, but subsequent studies (e.g. Brugam
1979, Battarbee 1979, Carney 1982, Charles 1985, Whitmore 1985)
have shown that the A:C index is not a useful indicator of lake
trophic status. The essential problem with these indices is that they
assume ecological uniformity of diatom species over broad taxonomic
groupings, whereas the individual species actually have ecologically
diverse requirements (C. Reimer pers. comm.).
Schelske et al. (1983) examined concentrations of biogenic silica
in sediment cores from the Great Lakes. Increases in biogenic silica
were shown over time in the sediments of all of the Great Lakes
because eutrophication led to a more rapid production and
sequestering to sediments of diatom valves. The peak in
sedimentary storage of biogenic silica in Lakes Ontario and Erie
occurred in the 1800's and was followed by a decline that resulted
from silica limitation (Kilham 1971) as these lakes continued to
eutrophicate. Sedimentary biogenic silica increased after 1940 in
Lake Michigan and reached a maximum abundance in 1964, after
which it declined. Stoermer et al. (1990) used cluster analysis to
delineate diatom zonation in a sediment core from Lake Michigan








and determined that changes in diatom species composition support
the eutrophication inferences of sedimentary biogenic silica.
Schelske (1988) demonstrated that recent declines in sedimentary
biogenic silica are consistent with historic water concentration data
that showed a decline in dissolved silica in Lake Michigan.
Whitmore (in press) studied the relationship in Florida lakes
between sedimentary diatom concentrations and accumulation rates
and lake trophic state as indicated by a TSI based on water-column
Chl a. Both periphyton and planktonic diatom concentrations were
positively correlated with water-column Chl a. Because diatom
accumulation rates were determined by three order of magnitude
differences in sedimentary diatom concentrations rather than by the
small range in bulk sediment accumulation rates, sedimentary
diatom concentrations were shown to be more expedient predictors
of Chl a than diatom accumulation rates. Sedimentary concentrations
were found to be unreliable predictors of trophic state when factors
such as silica limitation or blue-green bacterial inhibition limit

phytoplankton production, or when post-depositional changes affect
preservation of diatom valves.
Bailey and Davis (1978) used a multiple regression of diatom
taxa to predict water-column total P in a set of 19 lakes in Maine.
The best model explained 96% of the variance of total P in these
lakes, but contained only a few species of Fragilaria as independent
variables. Such models based on a limited number of taxa may
prove unreliable when applied to lakes outside of their calibration
data sets because of the large number of environmental factors that
can influence the distribution and abundance of species (Patrick








1973). Predictive approaches that utilize groups of taxa are generally
more reliable than those based on a limited number of taxa
(Battarbee 1979).
Diatom indices have been proposed that used trophic
autecological classifications of taxa for lakes in Florida (Whitmore
1985, 1989) and in Canada (Agbeti and Dickman 1989). In both of
these studies, diatoms were classified into 5 autecological categories,
and their percentages were structured into an index similar to
indices used for pH reconstructions (Nygaard 1956, Renberg and
Hellberg 1982). Log-transformed values of the indices were
regressed with log-transformed total P and Chl a in the Canadian
lakes and with TSI(TP) and TSI(Chl a) in Florida lakes. Log-
transformed values of the diatom inferred trophic index (D.I.T.I.)
(Agbeti and Dickman 1985) explained 71% of the variance in log-
transformed total P in the Canadian lakes, and the TROPH 1 index
(Whitmore 1989) explained 83% of the variance in TSI(TP) in Florida
lakes. Agbeti and Dickman concluded that the D.I.T.I. diatom index
was influenced by unspecified environmental factors. Whitmore
showed that pH was an important covariable affecting diatom
assemblages in the Florida lakes, though partial correlations
demonstrated that the predictive model using the TROPH 1 diatom
index was not statistically confounded by pH. Despite the fact that
silica limitation or cyanobacterial inhibition may affect
paleoproductivity inferences based on diatom accumulation rates
(Anderson 1990c), the TROPH 1 index still seems useful because
diatom assemblages are qualitatively distinct at the high nutrient
conditions where their populations are limited (Whitmore in press).








Canfield (pers. comm.) pointed out that while diatom indices such as
TROPH 1 may be accurate predictors of water-column total P, they
are not comprehensive indicators of lakeside trophic state because
they ignore the important component of primary production that is
in macrophytes.
Anderson et al. (1990) used the reciprocal averaging (RA)
indirect ordination method to study historical changes in lake trophic
state in Lough Augher, Northern Ireland resulting from loading and
later re-direction of nutrients from point sources. Anderson et al.
did not correlate RA axes with water quality indicators to
quantitatively assess trophic state changes, but instead plotted RA
axes against each other to graphically depict assemblage similarity
and ecological change.
Charles (1985) investigated the relationship between lake-water
characteristics and sedimentary diatom assemblages in 38
Adirondack lakes. Charles used reciprocal averaging to determine
which environmental variables influenced the diatom assemblages,
and found that total P was a weak correlate with the first RA axis.
He concluded that pH proved a more important determinant of
diatom assemblage composition in the Adirondack lakes because
those lakes spanned a wider range of pH than of trophic state.
Huttunen and Merilainen (1986) used detrended correspondence
analysis, another indirect ordination method, to interpret historical
limnological trends in a Finnish lake and were able to demonstrate
eutrophication following deforestation and the inception of
agriculture, as well as recent lake acidification.








Fritz (1990) used the canonical correspondence option of CANOCO
(ter Braak 1987) in a constrained ordination of 127 diatom taxa in 64
lakes of the northern Great Plains. The ordination axis was
constrained by the variable salinity, which resulted in a predictive
model that Fritz used in reconstructing historical changes in the
salinity of Devils Lake.
In recent studies of Canadian lakes (Christie and Smol 1990, Hall
and Smol 1990), attempts to construct trophic predictive models
have involved canonical correspondence analysis as an explanatory
ordination method to identify limnological variables affecting diatom
assemblages in lakes having a wide range of trophic state but a
narrow range of pH. Weighted averaging calibration (Line and Birks
1990) was then used as the regression method to construct transfer
functions and determine historical changes in trophic variables.
Anderson (1990c) examined the weighted averaging approach to
quantitative trophic-state reconstruction and warned that weighted
averaging studies largely utilize open water phytoplankton and v
chemical data, and that they ignore littoral community production
and chemistry. Anderson suggested that modeling methods should
be coupled with the use of multiple cores to calculate whole-basin
diatom accumulation rates that would give a reliable measure of both
plankton and periphytic paleoproduction.


Comments on Statistical Methods
Cluster analysis is a statistical method that groups observations
into clusters that reflect their similarity without a priori
consideration of which factors are influencing the similarities.








Cluster analysis of diatom assemblages from a set of lakes, for
instance, would result in clusters of diatom species that demonstrate
a similar response to the environmental variables responsible for
between-lake variance in the assemblages.
Indirect ordination techniques, which include reciprocal
averaging, principal components analysis and detrended
correspondence analysis, reduce the number of variables (e.g. diatom
taxa) by combining the variables into a series of linear combinations
of the original variables called ordination axes. With indirect
ordination methods, ordination axes are just particular combinations
of variables that are uncorrelated and appear in the order that best
explains the variance within the data set. The relationship between
indirect ordination axes and environmental variables that influence
the data set can then be determined by correlating axes with
environmental variables. Other ordination axes, however, may be
more suitable for establishing the relationship between the taxa and
specific environmental variables influencing diatom assemblages.
Canonical correspondence analysis (CCA), included in canonical
community ordination (CANOCO), is a direct ordination technique in
which variables are combined into ordination axes that are
constrained by specific environmental variables (ter Braak 1987).
Ordination axes are independent and uncorrelated, and are created in
order of their variance explained by the environmental variables.
CANOCO, therefore, effectively inserts a regression model into the
ordination model. When a single environmental variable is specified
in the CCA procedure, CANOCO can be used to obtain eigenvectors to








construct a calibration model for that environmental variable (ter
Braak 1987).
Principal components analysis is a linear ordination method in
which species demonstrate a linear response over the ordination axes
and species coefficients, called eigenvectors, are calculated as slopes
of those lines. In weighted averaging methods, which include
detrended correspondence analysis (DCA), species are assumed to
respond in a modal fashion over the ordination range, and
coefficients are equal to the center or optimum of their distribution
curve along the range of ordination values (ter Braak 1987). CANOCO
is an extension of DCA that also assumes a modal species distribution
over ordination axes (ter Braak 1987).
Reciprocal averaging (RA), or factor correspondence analysis, is
an extension of principal components. Hill and Gauch (1980)
compared RA and DCA and concluded that DCA was a better method.
A main fault they cite with RA is that the second ordination axis
demonstrates an 'arch effect' that is a mathematical artifact relating
to no real structure in the data. Charles (1985), for instance,
observed this arch effect in his study of diatom communities of
Adirondack lakes and found it made ecological interpretations
difficult. A second fault of RA is that it does not preserve ecological
distances between species along the ordination axes (Hill and Gauch
1980). Anderson et al. (1990) have presented, on the other hand, an
argument that RA is a preferred method over DCA because DCA
destroys spatial relationships between successive samples that is
necessary to demonstrate a time trajectory of community response in








sediment cores, and it interferes with assessment of the importance
of species on samples.


Effects of Spatial Variation on Diatom Assemblages
In paleolimnological work, there is frequently an implicit
assumption that diatom assemblages have been homogenized by
resuspension prior to deposition so that a single surface-sediment
sample or a sediment core reflects lakeside mean limnological
conditions. If spatial variability exists in species composition of
diatom assemblages in surficial sediments, variance is introduced
into the calibration data sets used for models describing
diatom/limnological relationships. Spatial variability in diatom
assemblages from sediment cores may also affect the precision of
historical inferences.
Anderson (1990a) studied variability in diatom concentrations
and accumulation rates in 10 sediment cores from Lough Augher and
found that diatom accumulation rates and concentrations were not

spatially uniform. Differences resulted partly from variance in bulk
sediment accumulation rates that was not related in a predictable
way to water depth (Anderson 1990b). Factors including localized
resuspension, stream inputs, slumping and the effects of
macrophytes on wind circulation patterns were responsible for the
spatial differences in sedimentation rates. Anderson concluded that
no single sediment core reflected the mean accumulation rate of the
whole basin.
Studies on the spatial heterogeneity of species composition in
surficial sediment samples have shown that no single sample








represents an "average" lakeside diatom assemblage (Earle et al.
1988). Variance due to site seems to increase when habitat
specificity of taxa is considered, i.e. when planktonic and periphytic
taxa are separated (Dixit and Evans 1986, Anderson 1990a).
Planktonic taxa typically show greater representation in deeper
water, whereas periphytic taxa show relatively little redistribution
and are more abundant in the littoral zone. With respect to spatial
variation within a lake basin, periphyton contribute a substantial
amount of the variance observed in diatom assemblages because
their substrate preferences lead to patchy distributions (Earle et al.
1988). In a sediment core from any particular site in a lake,
however, periphytic diatoms tend to demonstrate more even
accumulation rates than do planktonic taxa (Anderson 1990a,
1990b).
Spatial variation in subfossil diatom assemblages does not seem
to invalidate construction of calibration data sets using single
samples from each lake when lakes are sampled over a limnological
range. Earle et al. (1988) have shown that single-sample, between-
lake differences are high enough to indicate that the comparison of
diatom assemblages between lakes is valid if samples are retrieved
from deeper areas with gentle slopes rather than from steep-sloped
areas.
Spatial variation does not preclude meaningful application of
predictive models to historical samples provided that the effects of
sample variance on inferences are understood. Anderson (1990a)
showed that sediment cores retrieved from deeper water sites
consistently demonstrated greater resolution of historical changes in








limnology than cores taken in littoral areas. Although taxon
resolution varied with sediment core site, each core from Lough
Augher gave fundamentally the same record of eutrophication. Dixit
and Evans (1986) concluded that when time or financial constraints
are important, a sediment core from the deepest site on a lake will
provide a reliable indication of historical trends in pH. Because of
differences in pH inferences from various sites, however, Dixit and
Evans indicate that it is important to analyze several sediment cores
and demonstrate replicability for absolute inferences.


Purpose
The purpose of this study is to develop methods that would
permit quantitative assessment of historical macrophyte biomass in
lakes using sedimentary indicators. Macrophyte standing crop has
been documented to be high (Canfield et al. 1983a) in lakes such as
Fairview in Orange Co., Florida that appear oligotrophic and have
been used for calibration of diatom/trophic state models (Whitmore
1989). Conventional paleolimnological reconstructions of trophic state
have focused on inferring water-column nutrient concentrations
principally from planktonic diatom assemblages, but they have
ignored the often substantial component of primary production
occurring in macrophytes.
Historical inferences of lower water-column nutrient
concentrations obtained with existing diatom predictive models don't
necessarily indicate that lakes were formerly less productive. A
negative correlation has been shown between water-column nutrient
concentration or phytoplankton biomass as measured by Chl a and








macrophyte biomass (Canfield et al. 1983a, Canfield et al. 1984).
Historic samples indicating low water-column nutrient concentrations
may represent times of high macrophyte production, especially if the
cyclic changes in water level of Florida's shallow lakes (Deevey 1988)
promote a periodic lakeward expansion of macrophyte beds. If the
trophic trajectory of lake ecosystems over time is to be fully
understood, a more holistic consideration of historical trophic state is
required, one that includes the macrophyte component of production.
Conventional sedimentary indicators of macrophytes are not
appropriate for quantitative reconstructions for a variety of reasons:

1) pollen and seed production is species specific quantitatively and
is often absent in plants such as Hydrilla that largely undergo
vegetative reproduction. Models predicting historical
macrophyte standing crop from sedimentary pollen or seeds
would have to be calibrated for each individual species;
2) there are no known photosynthetic pigments that would be
preserved in sediments and that are specific to macrophytes;
3) diagenesis often affects the preservation of macrophyte remains.
Diatoms are considered as potential macrophyte indicators in
this study because diatoms are usually well-preserved in lake
sediments and they are ecologically specific. Life-form classifications
are available (Lowe 1974) that permit separate consideration of
planktonic and periphytic taxa. Sedimentary concentrations and
accumulation rates of periphytic taxa might be expected to
demonstrate a positive correlation with the amount of submerged
macrophyte biomass for 2 reasons. First, a positive relationship
exists between periphytic biomass and the increased substrate area
afforded by macrophytes with many small or finely-dissected leaves
such as Hydrilla (Cattaneo and Kalff 1981). Secondly, evidence also









indicates that diatom biomass might be stimulated by nutrient

release from macrophyte substrates (Burkholder and Wetzel 1990).
The specific objectives of this study are to:


1) obtain information on diatom assemblages in a calibration set of
Florida lakes that represent a wide range of macrophyte
presence;
2) identify periphyton and planktonic components of the
assemblages;
3) perform explanatory analyses to determine which variables
influence diatom assemblages over the range of macrophyte
presence;
4) construct predictive models that could be used to quantitatively
assess historical macrophyte standing crop using percentages,
sedimentary concentrations or accumulation rates of diatoms;
and
5) derive a plan to assimilate historic macrophyte inferences into a
scheme that permits a more complete assessment of former lake
trophic state than models restricted to concentrations of water
column nutrients.
I propose to develop new multivariate models predicting
historical macrophyte presence in the following manner. For

explanatory analyses of diatom communities, a cluster analysis will
be used to identify diatom taxa with similar ecological responses. An
indirect (unconstrained) ordination method, such as PCA, will be used
to ordinate taxa in linear combinations that best explain the variance
between assemblages. The ordination axes will then be correlated
with environmental variables to determine which variables exert the
greatest influence on diatom assemblages. Multivariate predictive
models will be constructed by stepwise linear regression of taxa and
by the direct ordination method canonical correspondence analysis,
in which the linear combinations of diatom taxa are constrained in a
manner best explained by the limnological variables of interest.













CHAPTER 2


METHODS


Collection of Sediment Samples
Sediment samples were collected from the sediment-water
interface of 30 Florida lakes. Lakes chosen for study were those
named in data sets (Canfield and Duarte 1988, Canfield unpub. data)
that contained data on macrophyte abundance. Lakes were selected
to cover the range of macrophyte presence as uniformly as possible.
Sediment samples were collected in two sets of field surveys. Survey
Set 1 consisted of samples from 10 lakes collected in the Spring of
1982. Samples were collected at a mid-lake station using an Ekman
dredge. Volumetric portions were removed with a pipette and
transferred to 125-ml Nalgene bottles. Sediment samples in Survey
Set 2 were collected between November 1987 and May 1988 from
20 additional Florida lakes. These samples were collected with a 77-
cm long, 8.8-cm diameter acrylic piston corer that was driven into
the sediment with 1 m long sections of magnesium-zirconium coring
rods. Water above the sediment-water interface was removed by
aspiration. The top 2 cm of sediment was collected with a syringe
and transferred to 125 ml Nalgene bottles.








Laboratory Analyses
Sediment samples were subdivided in the Paleoecology
Laboratory of the Florida Museum of Natural History for diatom
analyses, estimation of bulk sediment accumulation rates by 210pb
assay, and determination of percent organic matter.
Sediment samples for diatom analyses were cleaned using
hydrogen peroxide and potassium dichromate (Van der Werff 1955).
Digested samples were diluted with deionized water in 400-ml
beakers and settled overnight. The supernatant solutions were
removed by vacuum aspiration and the process of dilution, settling
and aspiration was repeated until the dichromate color was no longer
visible. Cleaned samples were suspended in 60 ml of water and
settled onto coverslips in evaporation trays (Battarbee 1973). Dried
coverslips were mounted on glass slides using Hyrax mounting
medium.
A minimum of 500 diatom valves was counted and identified on
a single slide of each sample using an American Optical Microstar
microscope at 1500X with a dark-phase condenser. Each diatom
valve was identified to the lowest taxonomic level possible using
standard diatom floras including those of Patrick and Reimer (1966-
1975) and Hustedt (1930,1930-1966).
The concentration of diatoms (D CONC) in the sediment samples
was calculated with the following formula:
D-CONC = C(PVD)-1

where D-CONC = number of frustules g-1 dry weight of sediment
C = number of valves counted
P = proportion of settling tray area counted







V = cm3 of sediment in initial sample
D = g dry weight of sediment cm-3 of initial sample.
Life-form autecological information for the majority of diatom
taxa was obtained from the literature (Lowe 1974, Patrick and
Reimer 1966-1975, Hustedt 1930). The concentrations of periphytic
valves (PERICONC) and euplanktonic valves (PLNKCONC) were
calculated by multiplying the concentration of diatoms in the
samples by the proportion of periphyton and euplankton in each
sample. Tychoplanktonic diatoms normally show a periphytic life-
form, though are often suspended with the plankton (Lowe 1974).
Tychoplanktonic taxa were arbitrarily assumed to be 1/3
euplanktonic for these calculations.
Bulk-sediment accumulation rates were measured for each
sediment sample using the Binford and Brenner (1986) dilution
tracer method based on 210Pb assay. 210Pb assay for 8 lakes from
Survey Set 1 was performed by a modification of the Eakins and
Morrison (1978) method that involved estimating 210Pb recovery of
samples from the proportion of recovery onto copper planchettes of a
208po spike. Atmospherically derived (unsupported) 210pb activity

was calculated as the difference between total residual 210pb activity
in the samples and an average supported 210pb activity (0.80 pCi g-1)
that was estimated from sediment cores of nine other Florida lakes
(Binford and Brenner 1986).
Sediment samples for 210pb assay of Survey Set 2 lakes were
dried, ground with mortar and pestle and weighed. These samples
were sealed in polypropylene tubes for greater than 3 weeks to
permit ingrowth of 226Ra daughter products. Radionuclide activity








was then measured using an ORTEC Intrinsic Germanium Detector
connected to a 4096-channel multichannel analyzer. The
unsupported activity was obtained by difference between total
residual 210Pb activity and supported activity of each sample as
assessed from 214Bi.
Bulk-sediment accumulation rates (SEDACCUM) were estimated
for samples from lakes in both survey sets using the following
formula:
SEDACCUM = F210PbA-1

where SEDACCUM = g dry sedimentcm-2 yr-1
F210pb = flux of 210pb fallout (pCi cm-2 yr-1)
A = unsupported 210Pb activity in sediment (pCi g-1).
Annual diatom accumulation rates (D-ACCUM) were estimated as
follows:
D-ACCUM = (D-CONCXSEDACCUM)
where D-ACCUM = valves cm-2 yr1.
The annual accumulation rates of periphytic valves (PERIACCM) and
euplanktonic valves (EUPLACCM) were obtained by multiplying the
periphyton and euplankton proportions of each sample by the total
diatom accumulation rates.
Subsamples of 1.001 cm3 wet sediment were dried at 60 oC in a
drying oven. Dry weight per unit volume (rho) was then measured
for each sample by weighing the dry material on an Ainsworth 24N
analytical pan balance. Percent loss on ignition (% L.O.I.), a measure
of sedimentary organic matter content, was determined for each
sample by weighing the sediment samples before and after





36


combustion in a muffle furnace for 1 hr at 550 OC. Percent loss on
ignition was then calculated as:


% L.O.I. = 100 x (1- (sample weight before combustion/sample
weight after combustion)).
Organic matter accumulation rates (ORGACCUM) were calculated with
the following equation:
ORGACCUM = SEDACCUM x (% L.O.I./100)
where ORGACCUM = g dry organic matter cm-2 yr-1.


Quantifying Macrophyte Presence
Researchers from the Department of Fisheries and Aquaculture
at the University of Florida obtained quantitative macrophyte data
for the survey lakes in field work conducted between 1982 and
1988. The percentage of the lake area covered by macrophytes
(percent-area coverage) was obtained for 23 lakes by planimetry on
morphometric maps. The area covered by submergent vegetation as
estimated from recording fathometer transects (Canfield and Duarte
1988) was added to the area covered by emergent and floating-
leaved vegetation. The percentage of the lake volume filled with
macrophytes (percent-volume infestation) was also estimated
morphometrically for 30 lakes using the percent areal coverage of
submerged macrophytes and the height of submerged vegetation as
indicated by fathometry (Maceina and Shireman 1980).
Morphometric data including lake surface area, shoreline length,
shoreline development and mean depth were available for 24 of the
survey lakes.








The average wet-weight biomass per unit area of submerged,
emergent and floating-leaved macrophytes was determined from
0.25 m2 quadrats randomly sampled from 10 transects through the
littoral zones of 23 lakes. Above-ground biomass in each quadrat
was collected by divers, spun to remove excess water and weighed to
the nearest 0.1 kg (Canfield and Duarte 1988).


Water Chemistry Data
Median water chemistry values for the lakes in Survey Set 1
were obtained from the Florida Lakes Data Base of the Water
Resources Research Center at the University of Florida. Water
quality variables selected from this data set included median values
for total P, total N, Chl a, Secchi depth, specific conductance and pH.
Mean water chemistry values were obtained by the Department
of Fisheries and Aquaculture for each lake in Survey Set 2 by
averaging data from 3 mid-lake stations (M. Hoyer, pers. comm.).
Total P analyses were performed following persulfate digestion
(Murphy and Riley 1962, Menzel and Corwin 1965), and total N was
measured using a modified Kjeldahl technique (Nelson and Sommers
1975). Water samples were filtered through a Gelman type A-E glass
fiber filter and Chl a was measured using the Yentsch and Menzel
(1963) method and Parson and Strickland (1963) equations. Secchi
depth was measured with a 20-cm black and white Secchi disk.
Trophic state index (TSI) values (Huber et al. 1982) were calculated
using the mean or median values of total P, total N, Chl a and Secchi
depth for each lake in Survey Sets 1 and 2. Specific conductance was
measured with a Yellow Springs Instrument Company model 31









conductivity bridge. pH was measured with an Orion 601A pH
meter.
Statistical Analyses
Data for diatom, macrophyte and water chemisty variables were
stored as Statistical Analysis System (SAS Inst., Inc. 1985) data sets
using the University of Florida's Northeast Regional Data Center. I
plotted the percent-area coverage, percent-volume infestation, and
submerged, emergent and floating-leaved biomass for macrophytes
against the percentage, concentration and accumulation rates of
diatom species to identify taxa responding to macrophyte presence.
Pearson product-moment correlation coefficients were obtained
between percent loss on ignition, organic-matter accumulation rates
and macrophyte variables using the SAS CORR procedure (SAS Inst.,
Inc. 1985). I also obtained correlation coefficients between
macrophyte variables and the percentages, concentrations and
accumulation rates of planktonic and periphytic diatoms. Correlation
coefficients were calculated between the macrophyte variables and
log-transformed concentrations and accumulation rates of diatoms.
In order to determine whether morphometric and chemical variables

have covariable effects on diatom-macrophyte relationships, I
obtained correlation coefficients between macrophyte, morphometric
and water chemistry variables.
For purposes of multivariate statistical analyses, it was essential
to reduce the number of diatom species from the 223 diatom taxa
observed. I examined plots of diatom percentages versus percent-
area coverage and percent-volume infestation, and selected forty-
seven taxonomic groups, several of which were the sum of taxa in








the same genus. Because of the large number of rare taxa present,
this preserved most of the diatom information in each sample (mean
= 96.6%) while substantially reducing the number of species.
Hierarchical cluster analysis was performed on the 47 taxonomic
groups using the SAS VARCLUS procedure (SAS Inst., Inc. 1985). I
applied this procedure to the percentages, concentrations and
accumulation rates of the taxonomic groups, and repeated the
procedures after partialling out the effects of TSI(AVG) and pH. Tree
diagrams of the hierarchical clusters from each analysis were
constructed using the SAS TREE procedure (SAS Inst., Inc. 1985).
Scores for diatom clusters were obtained for each lake in the survey
using standardized scoring coefficients from the cluster analyses and
the SAS SCORE procedure (SAS Inst., Inc. 1985). I then correlated the
scores for each cluster with macrophyte, chemical and morphometric
variables using the SAS CORR procedure to determine which
variables most influenced each cluster of taxa.
Principal components analyses of the percentages, concentrations
and accumulation rates of the 47 diatom taxonomic groups were
performed using the SAS PRINCOMP procedure (SAS Inst., Inc. 1985).
I repeated the PRINCOMP procedure for each of the three models
while partialling out the effects of TSI(AVG) and pH. The
standardized principal component scores of the first 8 principal
components in each test were calculated for the diatom assemblages
in the survey lakes using the SAS SCORE procedure. I then correlated
the principal component scores with macrophyte, water chemistry
and morphometric variables to identify the environmental variables
influencing each principal component.








Canonical correspondence analysis (CCA) was performed on the
percentage data for 47 diatom taxonomic groups using the canonical
community ordination (CANOCO) statistical package developed by ter
Braak (1987). In the first set of computations, the diatom groups
were ordinated in an axis constrained by the environmental variable
percent-volume infestation. In a second set of computations, 47
diatom taxa were ordinated into an axis constrained by percent-area
coverage.
Multivariate models that predict percent-volume infestation,
percent-area coverage and biomass for submerged, emergent and
floating-leaved plants were derived using the maximum R2
improvement method of the SAS STEPWISE procedure (SAS Inst., Inc.
1985). I reduced the number of independent diatom variables in
each stepwise regression to 20 or less, a recommended maximum for
this procedure (SAS Inst., Inc. 1985), after examining plots of diatom
taxonomic groups versus macrophyte variables. Regressions were
performed for each macrophyte variable using percentage data for
the diatom groups. The regressions for percent-area coverage and
percent-volume infestation were repeated using concentration and
accumulation rate data for the diatom groups.
I selected the best model in each STEPWISE regression
procedure by plotting Mallows' Cp statistic versus the number of
variables in each model (p) and selecting the model in which Cp was
approximately equal to p (Daniel and Wood 1971). Adjusted R2s,
which show the coefficient of determination after removing the
inflating effect of dependent variables, were calculated using the SAS
RSQUARE procedure (SAS Inst., Inc. 1985).








Partial correlations (Ott 1977) were used to determine if
predictive models were statistically free from confounding effects of
environmental variables correlated with their dependent variables.
I accomplished this by calculating multiple correlation coefficients as
the square root of the coefficients of determination for the
multivariate models. The multivariate models were then regressed
with the covariant dependent variables using the SAS GLM
procedure (SAS Inst., Inc. 1985), and I calculated multiple correlation
coefficients for these confounded forms of the model. Partial
correlation coefficients were calculated using the multiple correlation
coefficients to assess whether models with correct dependent
variables were significant when the effects of the covariables were
held constant.
Diatom-index values (TROPH 1) (Whitmore 1989) were
calculated for the subfossil diatom assemblages in the survey lakes.
A predictive model was developed to yield water-column total P
estimates from subfossil diatom assemblages for historic WCP
(Canfield et al. 1983a) inferences. The 51 lakes used to construct this
model were the lakes included in the present study, lakes in
Whitmore's (1989) study, and Lake Francis in Highlands County.
Water-column total P values for the lakes outside of the present
study were median values obtained from the Florida Lakes Data
Base. TROPH 1 values for subfossil diatom assemblages were
regressed with water-column total P values using the SAS GLM
procedure (SAS Inst., Inc. 1985).
Because of the significant negative correlation between
macrophyte presence and water-column nutrient concentrations





42


(Canfield et al. 1984), it was important to assess whether macrophyte
presence would have a significant confounding effect on the model
for predicting water-column total P. Log-transformed TROPH 1 index
values were correlated with log-transformed total P and percent-
area coverage values for the survey lakes.













CHAPTER 3


RESULTS


Thirty lakes were sampled in the surface-sediment survey
(Table 1). Lake Miona was removed from the data set prior to
statistical analyses because the lake was observed to contain
virtually no macrophytes despite a percent-volume infestation value
of 86% measured by Canfield (unpub. data). Triploid grass carp had
been introduced to Lake Miona by the Florida Game and Freshwater
Fish Commission in the time between the macrophyte and surface-
sediment surveys in order to control aquatic vegetation (M. Hoyer,
pers. comm.).
Many correlation coefficients were used in this study to assess
relationships between morphometric, water-chemistry and
macrophyte variables and to determine which environmental
variables influenced specific groups of diatoms. The criterion for
significance of all correlation coefficients discussed below is the a =

0.05 level of significance.
Water-chemistry values (Table 2a) showed that survey lakes
ranged from ultraoligotrophic to hypereutrophic (TSI(AVG): 4.1 to
88.5) as determined by water-column nutrient concentrations, and
from acidic to alkaline conditions (pH: 4.61 to 9.03). Total N/total P
ratios suggested that Lakes Wauberg and Alligator were N-limited
(Huber et al. 1982). Log-transformed water-column total P was










Table 1. Lake, county and sampling dates for macrophyte and
diatom surveys.



Lake County Macrophyte Diatom
date date


Alligator
Apopka
Bonny
Carr
Catherine
Clay
Crooked
Deep
Fairview
Harris
Hartridge
Keys Pond
Lindsey
Live Oak
Lochloosa
Loften Ponds
Miona
Moore
Mystic
Ocean Pond
Okahumpka
Orange
Patrick
Rowell
Stella
Tomohawk
Townsend
Watertown
Wauberg
Wildcat


Columbia
Orange
Polk
Leon
Marion
Lake
Lake
Putnam
Orange
Lake
Polk
Putnam
Hernando
Osceola
Alachua
Leon
Sumter
Leon
Madison
Baker
Sumter
Alachua
Polk
Bradford
Putnam
Marion
Lafayette
Columbia
Alachua
Lake


2 Jun
Sep
22 Sep
7 Jul
8 Sep
15 Jul
9 Jun
16 Jun
Oct
12 Oct
11 Aug
9 Jun
10 May
24 May
22 Aug
16 May
19 Aug
18 May
Jul
Aug
Aug
Oct
21 Jun
9 Aug
Sep
18 Jul
Jul
Aug
22 Jul
Aug


1987
1981
1987
1987
1987
1986
1987
1987
1982
1987
1987
1986
1988
1988
1988
1988
1986
1988
1982
1982
1981
1982
1988
1988
1981
1988
1981
1982
1986
1982


19 Dec
27 Jul
17 Dec
15 Jan
8 Dec
20 Jan
17 May
1 Feb
28 Jul
25 Mar
17 Dec
2 Feb
16 Nov
12 Jan
9 Nov
14 Jan
6 Jan
15 Jan
4 May
3 May
26 May
15 Aug
18 Dec
5 Dec
10 Aug
8 Dec
4 May
3 May
7 Nov
3 Aug


1988
1982
1988
1989
1988
1989
1989
1989
1982
1989
1989
1989
1988
1989
1988
1989
1989
1989
1982
1982
1982
1982
1988
1988
1982
1988
1982
1982
1988
1982









Table 2a.


Summary water chemistry data for survey lakes.
Sources of data were Canfield (unpub. data) and
Florida Lakes Data Base.


Lake Water-column TSI(AVG) TN/TP pH Specific
total P conductance
(mg 1-1) (gS/cm)


Alligator
Apopka
Bonny
Carr
Catherine
Clay
Crooked
Deep
Fairview
Harris
Hartridge
Keys Pond
Lindsey
Live Oak
Lochloosa
Loften Ponds
Moore
Mystic
Ocean Pond
Okahumpka
Orange
Patrick
Rowell
Stella
Tomohawk
Townsend
Watertown
Wauberg
Wildcat


0.320
0.192
0.050
0.015
0.003
0.001
0.007
0.002
0.015
0.028
0.010
0.002
0.017
0.014
0.032
0.004
0.005
0.015
0.040
0.020
0.040
0.010
0.069
0.010
0.004
0.009
0.062
0.158
0.008


74.9
88.5
71.2
42.4
16.6
7.4
24.3
4.1
27.4
66.4
33.6
6.7
41.2
35.3
61.0
20.5
19.7
25.7
46.0
47.2
59.4
35.5
66.0
25.1
15.7
29.5
51.7
77.1
20.9


7.4
21.0
37.2
58.0
100.0
360.0
47.1
80.0
33.3
55.4
48.0
85.0
38.2
25.0
32.8
97.5
70.0
34.7
10.5
47.5
27.8
147.0
11.7
43.0
52.5
64.4
16.9
9.9
23.9


7.9
8.0
7.7
6.3
4.7
4.8
4.6
4.6
8.1
8.5
7.8
5.4
6.8
7.0
7.7
4.8
5.8
6.7
5.0
9.0
7.2
8.1
7.7
7.1
4.9
5.2
7.4
7.4
4.8


144.0
395.0
255.8
25.3
48.3
52.0
44.3
37.0
198.8
247.7
219.3
42.7
34.0
131.0
87.7
19.0
16.3
28.0
45.5
201.7
82.5
320.3
289.7
240.0
34.7
23.4
153.3
80.0
33.0









Table 2b.


Summary data for macrophyte variables of survey lakes.
Source of data was Canfield (unpub. data).


Lake Percent Percent Floating Submerged Emergent
area volume -leaved biomass biomass
coverage infestation biomass
(kg m-2) (kg m-2) (kg m-2)


Alligator
Apopka
Bonny
Carr
Catherine
Clay
Crooked
Deep
Fairview
Harris
Hartridge
Keys Pond
Lindsey
Live Oak
Lochloosa
Loften Ponds
Moore
Mystic
Ocean Pond
Okahumpka
Orange
Patrick
Rowell
Stella
Tomohawk
Townsend
Watertown
Wauberg
Wildcat


10
3
10
100
48
100
27
97

27
60
40
100
100
83
87
40


100

93
43

43

7
0


10
0
7
100
9
76
2
21
33
2
11
8
80
55
57
22
14
78
0
95
79
42
10
39
12
65
1
1
2


1.2
1.1
0.0
7.0
1.1
4.5
3.8
2.5

0.8
0.0
0.0
1.3
0.3
0.6
0.6
0.2


8.8

0.4
0.3

0.5

0.0
11.2


0.0
0.0
3.0
9.9
2.9
6.8
2.4
11.7

0.9
8.0
1.0
1.8
1.6
2.6
0.7
1.3


16.6

1.3
0.0

1.0

0.0
4.4


1.7
2.5
8.1
12.7
4.6
8.1
26.8
10.6

2.4
4.9
2.8
3.0
2.0
2.2
0.3
1.7


11.9

1.1
0.4

1.4

1.0
12.1








Table 2c.


Summary morphometric data for survey lakes. Sources
of data were Canfield (unpub. data) and Florida
Lakes Data Base.


Lake Mean Lake Shoreline Shoreline
depth area length develop-
ment


Alligator
Apopka
Bonny
Carr
Catherine
Clay
Crooked
Deep
Fairview
Harris
Hartridge
Keys Pond
Lindsey
Live Oak
Lochloosa
Loften Ponds
Moore
Mystic
Ocean Pond
Okahumpka
Orange
Patrick
Rowell
Stella
Tomohawk
Townsend
Watertown
Wauberg
Wildcat


1.1
1.6
2.0
1.9
3.2
2.3
2.3
3.0

4.0
3.4
2.9
2.2
3.0
1.8
2.6
2.9



1.8
1.8
1.3

4.4
1.5
3.8
3.6
'


137
12412
143
254
41
5
8
4
163
5580
176
5
55
152
2309
5
28
19
722
394
5142
159
147

15
44
19
100
94


5.3
54.9
6.4
5.1
4.5
0.9
2.0
1.6

61.3
5.5
1.0
3.2
5.0
22.6
2.0
1.8




4.65
5.18

4.01
3.58
1.64
8.35


1.3
1.4
1.5

2.0
1.2
2.0
2.3

2.3
1.2
1.3
1.2
1.1
1.3
2.6





1.04
1.21

2.92
1.52
1.06
2.36








found to be highly correlated with specific conductance and pH
(Table 3, Fig. 1). Macrophyte variables were not significantly
correlated with water chemistry or morphometric variables, except
for percent-area coverage, that had a significant negative correlation
with TSI(AVG) (Fig. 2) and log-transformed total P (Table 3).
Two hundred twenty-three diatom taxa were found in the recent
sediments of the survey lakes (Appendix 1). Initially, 125 taxa had
life-form preferences described in the literature (Lowe 1974, Patrick
and Reimer 1966-1975), but the average percentage of valves with
unknown life-form preference was 15.04%. Life-form preferences
were assumed for 28 additional taxa (Appendix 1) based on valve
morphology. Taxa with a raphe that belonged to genera known to be
largely periphytic were assumed to have a periphytic life-form
preference. Two small species of Cyclotella and the long, lineate
Nitzschia romana and Synedra filiformis var. exilis were assumed to
be tychoplanktonic. These assumptions were discussed with Rex L.
Lowe (pers. comm.), who believed them to be correct. The new life-
form assignments reduced the mean percentage of valves with
unknown life-form preference in survey samples to 2.03%.
TSI(AVG) and pH were negatively correlated with the proportion
of periphytic diatoms (Fig. 3) and positively correlated with the
proportion of planktonic diatoms (Fig. 4) in surficial samples (Table
4). Diatom concentrations and accumulation rates were found to
vary as much as 4 orders of magnitude within the study lakes. Log-
transformed accumulation rates produced more significant
correlation coefficients with environmental variables than










Table 3. Pearson product-moment correlation coefficients between
water quality and macrophyte variables for survey
lakes. = p < 0.05.


correlation coefficient
prob. > 1 r 1 under Ho: rho = 0
sample size


Percent
area
coverage


Percent
volume
infestation

Floating-
leaved
biomass

Submerged
biomass


Emergent
biomass



pH


Specific
conductance


logl0
total P
*-0.558
0.007
22

-0.169
0.380
29

0.124
0.582
22

-0.252
0.258
22

-0.124
0.582
22

*0.679
<0.001
29

*0.493
0.007
29


TSI(AVG)

*-0.518
0.014
22


-0.131
0.497
29

0.134
0.552
22

-0.219
0.327
22

-0.121
0.592
22

*0.713
<0.001
29

*0.569
0.001
29


pH

-0.185
0.410
22

0.136
0.482
29

0.072
0.750
22

0.033
0.883
22

-0.253
0.256
22


Specific
conductance
-0.298
0.177
22


-0.212
0.269
29

-0.194
0.388
22

-0.136
0.546
22

-0.251
0.260


*0.755
<0.001
29









Table 3 cont'd.


correlation coefficient
prob. > 1 r 1 under Ho: rho = 0
sample size


Percent
area
coverage

Percent
volume
infestation

Floating-
leaved
biomass

Submerged
biomass


Emergent
biomass



pH


Specific
conductance


Lake
surface

-0.322
0.145
22

-0.106
0.593
28

-0.11
0.626
22

-0.2107
0.347
22

-0.165
0.462
22

0.343
0.074
28

*0.51
0.005
28


Mean
depth


-0.137
0.553
21

-0.396
0.062
23

0.043
0.852
21

0.0739
0.750
21

-0.035
0.880
21

-0.196
0.371
23

-0.216
0.322
23


Shoreline
length

-0.314
0.165
21

-0.230
0.304
22

-0.084
0.717
21

-0.238
0.299
21

-0.175
0.449
21

*0.508
0.016
22

*0.563
0.006
22


Shoreline
develop.

-0.173
0.452
21

-0.401
0.064
22

0.184
0.426
21

-0.007
0.976
21

0.168
0.467
21

-0.370
0.090
22

-0.241
0.281
22
















*

*
*
0 *e*@ S
** *
0 0
*
*
*
*

0*
0*~ 0
.* *..

r = 0.71


Figure 1.


9.5


8.5-


7.5-


6.5-


5.5-


4.5-


3.5-


100


for 29


S 40 60 80
TSI(AVG)
Plot of pH versus TSI(AVG)
lakes in synoptic survey.


20





52









100- r = -0.56
*
*
0 80-


2 60


40- *

0 *
20-
*


-3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0
Log 0 limnetic total P

Figure 2. Plot of percent area coverage
versus log-transformed limnetic
total P for 29 survey lakes.















1.00 r = -0.57

S 0.90

S 0.80- *

0.70
I-
0.60
0
0.50

0.40

0.30 -
0 20 40 60 80 100
TSI(AVG)
Figure 3. Proportion of diatom assemblage
that periphyton represent versus
TSI(AVG) for 29 lakes in survey.














0.6


0.5-


0.4-


0.3-


0.2-


0.1-


0.0


Figure 4.


40 60 80 100
TSI(AVG)
Proportion of diatom assemblage
that plankton represent versus
TSI(AVG) for 29 lakes in survey.


20
20


*
r = 0.57 *
0
*
0
0





*




** *
-* -------------0--








Table 4.


Correlation coefficients for diatom variables, which are
based on proportions, sedimentary concentrations and
annual accumulation rates of periphyton and plankton,
with water quality, macrophyte and morphometric
variables. Keys Pond was removed from periphyton
concentration data because of an anomalously high diatom
concentration value (see Fig. 5). = p < 0.05.


correlation coefficient
prob. > I r l under Ho: rho = 0
sample size


logio
total P



TSI(AVG)



pH


PERI-
PROP


*-0.610
<0.001
29

*-0.566
0.001
29

*-0.404
0.030
29


Specific -0.223
conductance 0.244
29


Percent-
area
coverage

Percent-
volume
infestation


*0.445
0.038
22

0.258
0.176
29


PLNK-
PROP

*0.613
<0.001
29

*0.573
0.001
29

*0.402
0.031
29

0.215
0.262
29

*-0.438
0.042
22

-0.243
0.204
29


PERI-
CONC

*0.410
0.030
28

*0.507
0.006
28

0.309
0.109
28

0.204
0.298
28

-0.220
0.339
21

-0.105
0.596
28


log
PLNK-
CONC

*0.644
<0.001
29

*0.656
<0.001
29

*0.492
0.007
29

*0.387
0.038
29

-0.353
0.107
22

-0.186
0.333
29


log
PERI-
ACCM

0.377
0.053
27

0.347
0.076
27

0.300
0.130
27

*0.392
0.041
27

-0.180
0.422
22

-0.280
0.158
27


log
PLNK-
ACCM

*0.630
<0.001
27

*0.578
0.002
27

*0.517
0.006
27

*0.480
0.011
27

-0.280
0.206
22

-0.203
0.311
27










Table 4


correlation coefficient
prob. > 1 r l under Ho: rho = 0
sample size


Floating-
leaved
biomass

Submerged
biomass


Emergent
biomass



Mean
depth


Shoreline
length


Shoreline
develop-
ment


PERI-
PROP

-0.097
0.668
22

0.193
0.391
22

0.165
0.463
22


-0.195
-0.372
22

-0.372
0.088
22

-0.002
0.992
22


PLNK-
PROP

0.126
0.576
22

-0.173
0.440
22

-0.143
0.525
22


0.169
0.440
22

0.371
0.089
22

-0.013
0.955
22


PERI-
CONC

0.221
0.336
21

-0.106
0.647
21

0.028
0.904
21


-0.096
0.669
22

0.131
0.572
21


-0.273
0.232
21


log
PLNK-
CONC

0.088
0.695
22

-0.140
0.534
22

-0.005
0.982
22


-0.059
0.789
22

0.337
0.125
22

-0.272
0.220
22


log
PERI-
ACCM

*-0.205
0.360
22

-0.311
0.159
22

-0.278
0.211
22


-0.246
0.270
22

0.155
0.492
22

-0.413
0.056
22


log
PLNK-
ACCM

0.019
0.932
22

-0.137
0.543
22

-0.113
0.616
22


-0.108
0.633
22

0.315
0.154
22

-0.330
0.133
22


cont'd.








untransformed values did. Table 4 shows that concentrations and
log-transformed accumulation rates of both periphyton (Fig. 5) and
plankton (Fig. 6) were positively correlated with TSI(AVG). pH, an
important correlate of TSI, was found to be negatively correlated
with the proportion of periphyton and positively correlated with the
proportion of plankton in recent sediments of survey lakes. pH was
also positively correlated with the log-transformed concentrations
and accumulation rates of plankton. Chl a, total N and Secchi depth
values used to calculate TSI(AVG) are shown in Appendix 4.
Appendix 6 lists the proportions, sedimentary concentrations and
accumulation rates of periphyton and plankton in the survey lakes.
The only macrophyte variable that demonstrated significant
correlation coefficients with diatom life-form variables was percent-
area coverage, which was positively correlated with the proportion of
periphyton, and negatively correlated with the proportion of
plankton (Table 4). Percent-area coverage was also negatively
correlated with the concentration of planktonic diatoms. Coefficients
of determination indicate that the 3 diatom life-form variables
correlated with percent-area coverage would each explain only 19-
24% of the variance in that variable. Percent loss on ignition and
organic matter accumulation rates (Table 5) were not significantly
correlated with any of the water chemistry or macrophyte variables.
Appendix 2 lists the 47 taxonomic groupings that were used in
multivariate analyses. Individual taxa were combined into groupings
based on taxonomic and ecological affinities, and selected for
multivariate analyses based on the results of plots of their

percentages versus macrophyte variables.
















*
(Keys Pond) r = 0.51






S

*
*
*



0
*
*g
0

*** *


I"


S I
20


I I
40 60
TSI(AVG)


Figure 5.


Concentration of periphytic
diatoms in surface sediments
versus TSI(AVG) for 29 lakes.


8-


7-


6-



,5
I4-


.o o


V-


80


100















r = 0.52















** *
0 *
~ 1
dlb


40


60


80


100


TSI(AVG)


Figure 6.


Concentration of planktonic
diatoms in surface sediments
versus TSI(AVG) for 29 lakes.


20
20









Table 5. Percent loss on ignition,
rates, and organic mater
in synoptic survey.


bulk sediment accumulation
accumulation rates for lakes


% Loss on bulk sediment Organic matter
Lake ignition accumulation accumulation
@ 550 OC (g cm-2 yr-1) (g cm-2 yr-1)


Alligator
Apopka
Bonny
Carr
Catherine
Clay
Crooked
Deep
Fairview
Harris
Hartridge
Keys Pond
Lindsey
Live Oak
Lochloosa
Loften Ponds
Moore
Mystic
Ocean Pond
Okahumpka
Orange
Patrick
Rowell
Stella
Tomohawk
Townsend
Watertown
Wauberg
Wildcat


59.14
64.10
53.46
26.53
55.21
43.98
50.11
69.65
8.20
80.93
53.56
30.49
17.12
86.06
36.10
33.60
17.21

39.00
67.10

44.30
69.00
14.00
63.00
79.60
60.30
72.40


0.07
0.09
0.03
0.02
0.03
0.03
0.02
0.10
0.22
0.04
0.05
0.02
0.03
0.16
0.04
0.02
0.04

0.10
0.06

0.25
0.31
0.17
0.03
0.04
0.07
0.04


0.04
0.06
0.02
0.01
0.02
0.01
0.01
0.07
0.02
0.03
0.03
0.01
0.01
0.13
0.01
0.01
0.01

0.04
0.04

0.11
0.21
0.02
0.02
0.03
0.04
0.03








Results of Cluster Analyses
Cluster analysis based on the percentage data for 47 taxonomic
groups produced 16 clusters of taxa. Nine of these clusters
demonstrated significant correlation coefficients with TSI(AVG), and
six clusters showed significant correlations with pH (Table 6). Four
clusters that were correlated with TSI(AVG) and pH also showed
significant correlations with specific conductance. Three clusters
were correlated with both lake surface area and shoreline length.
Mean depth and shoreline development were each correlated with
single separate clusters. Of the 80 correlation coefficients between
diatom clusters and macrophyte variables, only two were significant,
one with percent-volume infestation, and one with emergent
biomass. Navicula pupula vars. (S-NAVPU) and Stauroneis spp. (S-
STAU) were the taxonomic groups in the cluster correlated with
percent-volume infestation, and these were included among the pool
of independent variables (Appendix 3.1) in a stepwise procedure to
construct a model that predicts percent-volume infestation. Caloneis
sp. A and Navicula seminulum vars. (S-NAVSEM) were the taxonomic

groups correlated with emergent biomass, but this correlation was
driven by a single datum because both species showed anamalously
high percentages in Crooked Lake.
Due to the overwhelming influence TSI(AVG) and pH had on
composition of taxonomic clusters, cluster analysis using taxonomic
percentages was repeated while partialling out the effects of pH and
TSI(AVG). Fourteen diatom clusters resulted from this procedure.
Seven of the clusters still demonstrated a significant correlation with
TSI(AVG), and six of these had significant correlations with pH. Two








Table 6. Pearson product-moment correlation coefficients for
clusters based on percentage data for diatom taxonomic
groups with macrophyte and water quality variables.
= p < 0.05.


correlation coefficient
prob. > 1 r 1 under Ho: rho = 0
sample size


TSI(AVG)



pH


Specific
conduc-
tance

Percent-
area
coverage

Percent-
volume
infestation

Floating-
leaved
biomass

Submerged
biomass


Emergent
biomass


CLUS1

*-0.691
<0.001
29

*-0.692
<0.001
29

*-0.531
0.003
29

0.145
0.519
22

-0.124
0.521
29

-0.149
0.508
22

-0.024
0.916
22

-0.045
0.843
22


CLUS2

0.163
0.397
29

*0.322
0.009
29

0.212
0.269
29

-0.084
0.710
22

-0.083
0.671
29

-0.116
0.607
22

0.121
0.591
22

-0.096
0.670
22


CLUS3

*0.547
0.002
29

*0.409
0.0276
29

0.218
0.257
29

-0.358
0.102
22

-0.194
0.313
29

-0.092
0.685
22

-0.279
0.209
22

-0.233
0.297
22


CLUS4 CLUS5


-0.173
0.368
29

-0.356
0.058
29

-0.253
0.185
29

0.193
0.388
22

-0.165
0.392
29

-0.019
0.934
22

-0.113
0.617
22

0.404
0.062
22


*0.528
0.003
29

*0.389
0.037
29

0.154
0.426
29

-0.271
0.215
22

-0.120
0.536
29

0.408
0.060
22

-0.106
0.640
22

0.040
0.859
22








Table 6 cont'd.


correlation coefficient
prob. > 1 R 1 under Ho: rho = 0
sample size


TSI(AVG)



pH


Specific
conduc-
tance

Percent-
area
coverage

Percent-
volume
infestation

Floating-
leaved
biomass

Submerged
biomass


Emergent
biomass


CLUS6

-0.081
0.677
29

-0.124
0.521
29

-0.150
0.437
29

-0.058
0.797
22

-0.009
0.963
29

0.1768
0.431
22

0.059
0.793
22

*0.724
<0.001
22


CLUS7

*0.485
0.008
29

*0.545
0.002
29

*0.415
0.025
29

-0.006
0.980
22

0.276
0.147
29

0.213
0.341
22

0.362
0.097
22

0.0731
0.746
22


CLUS8 CLUS9 CLUS10


*0.387
0.038
29

0.304
0.109
29

0.209
0.276
29

-0.296
0.180
22

-0.151
0.435
29

-0.225
0.312
22

-0.190
0.396
22

-0.251
0.259
22


-0.101
0.603
29

*0.381
0.041
29

*0.392
0.036
29

0.226
0.312
22

0.107
0.582
29

-0.249
0.262
22

-0.097
0.666
22

-0.109
0.626
22


-0.044
0.819
29

0.250
0.191
29

0.335
0.075
29

-0.018
0.935
22

0.018
0.927
29

-0.103
0.645
22

-0.225
0.313
22

-0.259
0.245
22








Table 6 cont'd.


correlation coefficient
prob. > 1 R I under Ho: rho = 0
sample size


CLUS11 CLUS12 CLUS13 CLUS14 CLUS15 CLUS16


TSI(AVG)


pH


Specific
conduc-
tance

Percent-
area
coverage

Percent-
volume
infestation

Floating-
leaved
biomass

Submerged
biomass


Emergent
biomass


*-0.471
0.009
29

*-0.515
0.004
29

-0.310
0.100
29

0.097
0.667
22

-0.185
0.335
29

-0.099
0.659
22

0.140
0.532
22

-0.011
0.958
22


0.006
0.971
29

0.112
0.559
29

-0.208
0.278
29

0.246
0.268
22

*0.439
0.017
29

-0.015
0.947
22

-0.043
0.847
22

0.021
0.925
22


*0.413
0.025
29

0.318
0.092
29

0.201
0.295
29

*-0.462
0.030
22

-0.227
0.234
29

0.004
0.985
22

-0.001
0.995
22

0.044
0.843
22


0.205
0.284
29

0.142
0.459
29

0.363
0.052
29

-0.190
0.396
22

-0.096
0.618
29

0.070
0.755
22

0.080
0.722
22

-0.038
0.864
22


*-0.4032
0.0301
29

-0.335
0.075
29

*-0.398
0.032
29

0.303
0.170
22


0.341
0.063
29

0.227
0.309
22

0.262
0.237
22

0.363
0.096
22


*-0.368
0.048
29

-0.330
0.079
29

-0.331
0.078
29

0.334
0.128
22

0.265
0.163
29

-0.175
0.435
22

0.008
0.968
22

-0.003
0.989
22








of the three clusters that were significantly correlated with specific
conductance were also correlated with TSI(AVG) and pH. Among the
morphometeric variables, two clusters were significantly correlated
with mean depth, and one cluster was correlated with shoreline
development. Two clusters were significantly correlated with both
lake surface area and shoreline length. Of the 70 correlation
coefficients between macrophyte variables and diatom clusters, only
three correlation coefficients were significant. One cluster that was
composed of Caloneis sp. A and Navicula seminulum vars.
demonstrated a significant correlation with emergent biomass as in
the unpartialled analysis, and this again appeared spurious because
the correlation was driven by anomalously high percentages of both
taxa in Crooked Lake. The second cluster, which consisted of
Anomoeneis spp. (S-ANOM), Gomphonema spp. (S-GOMA), Navicula
pupula vars., and Stauroneis spp. (S-STAU), was correlated with both
percent-area coverage and percent-volume infestation. Taxonomic
groups in this cluster that showed a directional change in abundance
over the range of percent-area coverage and percent-volume
infestation were included among the independent variables in the
stepwise regression procedures to predict these macrophyte
variables (Appendix 3.1 and 3.4).
Cluster analysis based on sedimentary concentrations of diatoms
produced 12 clusters, 4 of which were significantly correlated with
TSI(AVG), and 5 of which were correlated with pH. The
morphometeric variables lake-surface area and shoreline length
were each significantly correlated with 2 separate clusters. Of 60
correlation coefficients between diatom clusters and macrophyte








variables, 4 were significant. A cluster that was composed of the
taxon Caloneis sp. A was significantly correlated with emergent
biomass, but this correlation was driven by a single datum point
because of the unusually high percentage of this taxon in Crooked
Lake. Another cluster consisted of Fragilaria brevistriata and
Gomphonema spp. (S-GOMA) and was significantly correlated with
percent-volume infestation. This correlation did not appear
meaningful because F. brevistriata was represented by only two data
points, and G. spp. was correlated with percent-volume infestation
only because of anamalously high values in Lake Carr. In addition, F.
brevistriata and G spp. showed different slopes over the range of
percent-volume infestation. A third cluster, which seemed
spuriously correlated with emergent biomass, consisted of two
euplanktonic and two periphytic taxa. Asterionella spp. (S-AST) and
Aulacoseira islandica were the two euplanktonic taxa in this cluster
and they showed high values only in Crooked Lake. Tabellaria
flocculosa and T. fenestrata were the periphytic taxa in this cluster
and their slopes were different over the range of emergent biomass.
A fourth cluster was positively correlated with floating-leaved
biomass (r = 0.554, p = 0.008, n = 22) and contained 9 taxonomic
groups (Appendix 3.7), seven of which were euplanktonic. These
taxa were used as independent variables in a stepwise regression
procedure to predict floating-leaved biomass.
When the clustering procedure was repeated partialling out the
effects of TSI(AVG) and pH, a cluster composed of the 7 euplanktonic
taxa last mentioned was again positively correlated with floating-
leaved biomass. Four clusters were still significantly correlated with








TSI(AVG) and 5 were correlated with pH. Eunotia spp. (S-EUN) and
Gomphonema spp. (S-GOMA) composed a cluster that was
significantly correlated with percent-volume infestation, but plots
revealed that both of these taxonomic groups had low sedimentary
concentrations except for unusually high concentrations in Lake Carr.
Emergent biomass was significantly correlated with a cluster
composed of Asterionella spp., Aulacoseira islandica, Tabellaria
flocculosa and T. fenestrata as it was prior to partialling covariant
effects. Another cluster composed of Caloneis sp. A and Nitzschia
capitellata was correlated with emergent biomass, but these taxa
demonstrated opposite slopes over the range of emergent biomass.
None of the 66 remaining correlation coefficients between diatom
clusters and macrophyte variables was significant.
Cluster analysis of diatom taxonomic groups based on annual
diatom accumulation rates produced 11 clusters. Four of these
clusters were significantly correlated with TSI(AVG), and three of
these plus a fourth cluster were significantly correlated with pH. Four
clusters were correlated with shoreline length and two clusters were
correlated with specific conductance. Of the 55 correlation
coefficients between diatom clusters and macrophyte variables, only
one correlation coefficient was significant. Floating-leaved biomass
was found to be negatively correlated with a cluster composed of 4
periphytic and 3 planktonic taxonomic groups (Appendix 3.8). These
diatom groups were used in a stepwise regression procedure to
predict floating-leaved biomass.
The last cluster analysis was based on diatom accumulation rates
with the effects of TSI(AVG) and pH partialled out, and it produced








12 clusters of diatom taxa. TSI(AVG), pH and specific conductance
had 3 significant correlations each with diatom clusters. Shoreline
development and lake surface area were correlated with 2 clusters
each, and mean depth was correlated with one cluster. Of 60
correlation coefficients related to the macrophyte variables, only two
were statistically significant. Seven of the 9 taxa in one cluster that
was correlated with floating-leaved biomass were the same taxa
used in the stepwise multiple regression procedure described above
for unpartialled effects of TSI(AVG) and pH. Another cluster
consisted of Fragilaria crotonensis, a euplanktonic taxon occurring in
lakes high in water-column nutrients, that had an apparently
spurious correlation with floating-leaved biomass.


Results of Principal Components Analyses
Principal components analysis seems to be an appropriate
indirect ordination method to apply because species distributions
were observed to be linear or curvilinear, though not modal, over the
range of percent-area coverage and percent-volume infestation.
Correlation coefficients were examined between the first 8 principal
components based on percentage data for 47 diatom taxonomic
groups and environmental variables. Eigenvalues, which are equal to
the variances of the components, indicate that the first 3 principal
components account for 15.9%, 8.5% and 7.2% of the variance in
diatom assemblages, respectively. The first principal component was
found to be highly correlated with TSI(AVG), pH and specific
conductance (Table 7), indicating that these environmental variables
were responsible for most of the variance in the diatom assemblages.








Table 7. Correlation coefficients for principle components based on
diatom percentage data with water quality and
macrophyte variables. = p < 0.05.


correlation coefficient
prob. > 1 r l under Ho: rho = 0
sample size

1st. 2nd. 3rd. 1st.
principle principle principle principle
comp. comp. comp. comp.
TSI(AVG) and
pH partialled out


Percent -0.556 0.130 *0.395 -0.134
volume 0.774 0.500 0.034 0.653
infestation 29 29 29 29

Percent -0.322 0.132 0.298 -0.171
area 0.144 0.557 0.176 0.444
coverage 22 22 22 22

Floating- 0.102 0.064 -0.146 0.086
leaved 0.649 0.776 0.515 0.703
biomass 22 22 22 22

Submerged -0.108 -0.163 0.168 -0.185
biomass 0.632 0.467 0.453 0.409
22 22 22 22

Emergent -0.144 0.114 0.082 -0.081
biomass 0.521 0.612 0.714 0.720
22 22 22 22

*0.753 -0.023 0.029 *0.414
TSI(AVG) <0.001 0.901 0.880 0.026
29 29 29 29








Table 7 cont'd.


correlation coefficient
prob. > I r 1 under Ho: rho = 0
sample size


1st.
principle
comp.


2nd.
principle
comp.


3rd. 1st.
principle principle
comp. comp.
TSI(AVG) and
pH partialled out


pH


Specific
conductance


Mean
depth


Shoreline
length


Lake
surface
area


*0.775
<0.001
29

*0.539
0.003
29

-0.170
0.437
23

0.200
0.369
22

0.150
0.445
28


-0.132
0.494
29

-0.283
0.135
29

*-0.437
0.037
23

*-0.457
0.032
22

*-0.454
0.015
28


*0.374
0.045
29

*0.373
0.045
29

-0.154
0.481
23

0.446
0.037
22

0.314
0.103
28


*0.411
0.026
29

0.187
0.330
29

-0.105
0.633
23

-0.192
0.390
22

-0.260
0.180
28








The second principal component was significantly correlated
with with the morphometric variables mean depth, shoreline length
and lake surface area. The third principal component was
significantly correlated with pH, specific conductance and percent-
volume infestation. A coefficient of determination indicates that the
third principal component would explain 16.0% of the variance in
percent-volume infestation, which is not sufficiently robust for
predictive purposes. None of the remaining 60 correlation
coefficients between principal components 4-8 and the
environmental variables was statistically significant.
Principal components analysis of percentage data was repeated
while TSI(AVG) and pH were partialled out. The first principal
component, which accounted for 11.8% of the variance in diatom taxa
still showed significant correlations with TSI(AVG) and pH (Table 7).
The second principal component explained 9.5% of the variance in
the diatom assemblages and was significantly correlated with mean
depth. The third principal component explained 8.6% of the variance
in the diatom assemblages and was not correlated with any of the
macrophyte or environmental variables considered. The fourth
principal component explained 7.5% of the variance, and was
negatively correlated with floating-leaved biomass (r = -0.442, p =
0.040, n = 22). A model based on this principal component would
explain approximately 19.5% of the variance in floating-leaved
biomass. None of the correlation coefficients between principal
components 5-8 and the environmental variables was significant.
Principal components analysis was performed on sedimentary
diatom concentrations for the 47 diatom taxonomic groups. The first








principal component explained 18.3% of the variance and was
significantly correlated with TSI(AVG) (r = 0.648, p < 0.001, n = 29)
and pH (r = 0.499, p = 0.006, n = 29). The second principal component
explained 10.3% of the variance in the diatom assemblages and was
not significantly correlated with any of the environmental or
macrophyte variables. The third principal component explained 9.9%
of the variance in the diatom assemblages. This component had a
significant negative correlation with floating-leaved biomass (r =
0.536, p = 0.010, n = 22), and a positive correlation with pH (r =
0.400, p = 0.032, n = 29) and shoreline length. Scores obtained for
survey lakes using eigenvectors of the third principal component
(Table 8) were used to construct the following model:
floating-leaved biomass = 2.419 0.750(PRIN3) 3.1
R2 = 0.287, p = 0.010, n = 22
where PRIN3 is the sum of the products between eigenvectors and
sedimentary concentrations of the 47 diatom groups.
The majority of the taxa in Table 8 that show large, positive
eigenvectors (e.g. Fragilaria construens, F. pinnata, Navicula
lanceolata, N. pupula and vars., N. radiosa and vars., N. cuspidata,
Nitzschia amphibia, N. capitellata, Cocconeis placentula var. lineata),
have a periphytic life form. Many of the taxa with smaller, negative
eigenvectors (e.g. Asterionella spp., Aulacoseira islandica,
Cyclostephanos dubius, Fragilaria crotonensis) have a euplanktonic
life form. It appears that samples with large numbers of periphytic
diatoms and few planktonic diatoms would have large values of
PRIN3, and it would be reasonable to expect that large PRIN3 values
would be associated with greater floating-leaved biomass. Equation









Table 8. Eigenvectors of third principle component
sedimentary concentrations of 47 diatom
groups. Taxonomic acronyms are defined


based on
taxonomic
in Appendix 2.


Taxonomic Eigenvector Taxonomic Eigenvector
group group


S-ACH
S-ANOM
S-AST
AULAAM
AULADIS
S-AULAGR
AULAISL
AULAITAL
CALSPA
COCPLACL
CYCMEN
CYCPSEUD
CYCSIEL
CYCSTELO
S-CYM
CYSTEPDU
S-EP
S-EUN
ACIPUNC
NAVCUS
NAVCONF
TABFLOC
S-NEI
FRAGBREV
S-FRAGCO
FRAGCROT
FRAGPIN
S-FRUSRH
S-GOMA


0.081
0.089
-0.177
0.186
-0.025
-0.096
-0.141
0.039
-0.013
0.138
0.124
-0.023
0.013
0.005
0.034
-0.177
0.035
0.007
-0.075
0.137
0.115
-0.168
0.103
0.069
0.298
-0.166
0.225
0.096
0.006


NAVGOT
NAVLAN
S-NAVPU
S-NAVRA
S-NAVSEM
NAVSUBT
NITZAM
NITZCAP
NITZFONT
NITZFRUS
NITZPAL
S-PIN
S-STAU
S-SUR
SYNDEL
SYNFILEX
S-SYNRUM
TABFEN


0.061
0.261
0.209
0.320
0.192
0.122
0.235
0.195
-0.139
0.193
-0.128
0.109
0.101
-0.037
0.269
0.123
-0.014
-0.098








3.1 shows a negative coefficient for PRIN3, however, which is
contrary to the expectation that periphytic taxa would demonstrate
greater representation in lakes with greater floating-leaved biomass.
Partial correlation coefficients showed that the 3rd principal
component was significantly correlated with floating-leaved biomass
(r = -0.618, p = 0.003, n = 22) when the effect of pH was held
constant. Despite the non-significant correlation between pH and
floating-leaved biomass (Table 3), partial correlations showed that
pH was significantly correlated with the 3rd principal component (r =
0.521, p = 0.016, n = 29) when floating-leaved biomass was held
constant. This implies that the model predicting floating-leaved
biomass is confounded by the effect of pH, and should only be
applied historically when it can be demonstrated that significant
changes in pH have not occurred.
When the principal components analysis based on the
sedimentary concentration of diatom groups was repeated partialling
out the effects of TSI(AVG) and pH, the first principal component
explained 16.0% of the variance in the diatom assemblages and was
positively correlated with floating-leaved biomass (r = 0.460, p =
0.031, n = 22) and with TSI(AVG). This correlation with floating-
leaved biomass was less robust than in the previous correlation with
unpartialled effects. None of the remaining principal components in
the partialled analysis had significant correlations with macrophyte
variables.
The principal components procedure was repeated using log-
transformed sedimentary diatom concentrations. The first principal
component accounted for 30.4% of the variance in the diatom








assemblage and demonstrated stronger correlations with TSI(AVG) (r
= 0.783, p < 0.001, n = 29) and pH (r = 0.885, p < 0.001, n = 29) than
the procedure with untransformed concentration data. None of the
correlation coefficients between principal components and
macrophyte variables were significant in this procedure, even when
the effects of TSI(AVG) and pH were partialled out.
Principal components analysis using annual diatom accumulation
rates produced a first principal component that explained 21.4% of
the variance and was again significantly correlated with TSI(AVG) (r
= 0.460, p = 0.012, n = 29), pH (r = 0.464, p = 0.011, n = 29), and
specific conductance (r = 0.501, p = 0.006, n = 29). The only
significant correlation coefficient involving a macrophyte variable
was with the fourth principal component that was significantly
correlated with percent-area coverage (r = -0.448, p = 0.036. n = 22)
and more strongly correlated with TSI(AVG) (r = 0.501, p = 0.006, n =
29). A predictive model using eigenvectors of this principal
component would not be useful for predicting percent-area coverage
because it would be confounded by TSI(AVG).
Principal components based on diatom accumulation rates with
the effects of TSI(AVG) and pH partialled out produced a single
significant correlation with a macrophyte variable. The sixth
principal component, which accounted for 5.5% of the variance in the
diatom assemblages, had a significant negative correlation with
floating-leaved biomass (r = -0.574, p = 0.005, n = 22). A predictive
model using the eigenvectors for this principal component would
explain 33.0% of the variance in floating-leaved biomass but could








only be applied historically to lakes that had not undergone changes
in trophic state or pH.


Results of Stepwise Multiple Regression
Percent-Volume Infestation
The maximum R2 method of the SAS STEPWISE procedure (SAS
Inst., Inc. 1985) was applied to percentage data for 17 diatom
taxonomic groups selected from plots of their abundance versus
percent-volume infestation (Appendix 3.1). The plot of Mallows' Cp
statistic versus the number of variables in each model is shown in
Figure 7. Models with larger Cp values have larger total error than
models with smaller Cp values (Daniel and Wood 1971). Models in
which Cp is larger than the number of independent variables plus
the intercept (p) are subject to bias error. Models with Cp values less
than p are subject to random errors. Figure 7 shows that the model
for predicting percent-volume infestation that consisted of one
diatom taxon (p = 2) showed substantial bias. All other multivariate
models predicting percent-volume infestation from percentage data
were subject to random error.
Stepwise regression to predict percent-volume infestation was
attempted by using sedimentary concentration data for 17 diatom
taxonomic groups (Appendix 3.2). All models demonstrated random
error. Better results were obtained when log-transformed
sedimentary concentrations were used in the stepwise procedure.
Figure 8 is the plot of Cp versus p for models using log-transformed
diatom concentrations. The model using 2 diatom taxonomic groups
(p = 3) shows bias. The model using 3 taxonomic groups (p = 4)





77






20


15-


10-


5-


0


-5. -- i -- i -- -- i -- -
0 5 10 15 20
P
Figure 7. Mallows' Cp statistic versus p for
model predicting percent volume
infestation from percentage data.
P is number of dependent variables + 1.



























0- I


-5 -* --i -i -


5 10 15


Figure 8.


Mallows' Cp statistic versus p for
model predicting percent volume
infestation from log-transformed
diatom concentrations.


20


15


10


20








shows a slight random error, but a lower total error term. The best
model for predicting percent-volume infestation (PVI) from
sedimentary concentrations of diatom groups is therefore:

PVI = 36.960 4.2171og(ACHEX) + 7.8331og(STAUPH) -
3.6411og(SYNFILEX) 3.2
R2 = 0.519, p < 0.001, n = 29
where values for the 3 taxa, whose acronyms are defined in
Appendix 1, are expressed in valves g-1 dry weight of sediment.
The adjusted R2 value for this model is 0.461.
The stepwise procedure was applied to log-transformed annual
accumulation rate values for 17 diatom taxonomic groups (Appendix
3.3). The best model utilized 2 diatom taxa and showed an R2 of
0.312. The adjusted R2 for this model was 0.259. The model was
less robust than the model based on log-transformed sedimentary
diatom concentrations.


Percent-Area Coverage
Eleven diatom taxonomic groups (Appendix 3.4) selected from
plots were used in the stepwise procedure to construct models
predicting percent-area coverage from diatom percentage data. The
best model, as determined from Cp values was:

percent-area coverage = 27.294 + 3.971(S-CYM) + 0.871(S-
FRUSRH) + 50.967(STAUPH) 3.3
R2 = 0.502, p = 0.005, n = 22
where values for the 3 taxonomic groups, which are defined in
Appendices 1 and 2, are expressed as a percentage of each diatom








assemblage. The adjusted R2 for this model was 0.419. This
multivariate model was significantly confounded with TSI(AVG) (R2 =
0.665, p < 0.001) and would therefore be unsuitable for predictive
purposes.
The stepwise procedure was applied to sedimentary diatom
concentration data for 8 taxonomic groups (Appendix 3.5). Cp
statistics indicated that the best model was;

PAC = 44.408 + 2.14 x 10-6(ACHLIN) 2.7 x 10-7(S-AULAGR) +
9.53 x 10-5(EUNINC) + 1.91 x 10-5 (STAUPH) 3.4

R2 = 0.607, p = 0.002, n = 22, adj. R2 = 0.514.
Acronyms for taxonomic groups are defined in Appendices 1 and 2.
The best model produced by the stepwise procedure using log-
transformed sedimentary diatom concentration data showed an R2 =
0.422, that was less robust than the model based on untransformed
data.
The stepwise procedure was applied to annual diatom
accumulation rates for 11 diatom taxonomic groups (Appendix 3.6).
The procedure was also used for log-transformed accumulation rate
data of these taxa. The best model included 3 diatom taxonomic
groups and showed an R2 = 0.415 (p = 0.006). This model bore a
stronger relationship with TSI(AVG) (R2 = 0.480, p < 0.001), and was
therefore significantly confounded by this trophic state variable.


Floating-Leaved Biomass
Stepwise multiple regression was performed on sedimentary
concentration and log-transformed sedimentary concentration of








valves of 9 diatom taxonomic groups (Appendix 3.7) that were
correlated with floating-leaved biomass in the cluster analysis
procedure. All models that resulted from these stepwise attempts
demonstrated random errors. Stepwise regression was also
performed on the 7 diatom taxa in the annual-accumulation rate
cluster analysis that were correlated with floating-leaved biomass.
The best model, as indicated by the Cp statistic, included a single
diatom taxon, and this model explained only 28.7% of the variance in
floating-leaved biomass.
Better results were obtained by stepwise regression of 12
diatom taxonomic groups (Appendix 3.9) that were selected from
plots of percentage data for diatom taxa versus floating-leaved
biomass. The best model obtained was the following:

FLOATING = 5.105 3.4551og1O(ACHS) 2.4931og 10(CYMMUEL)
+ 0.264(CYSTEPDU) 0.281(S-NAVS) 3.2971oglO(S-
NITZS) 3.5
R2 = 0.866, p < 0.001, n = 22, adj. R2 = 0.825


where FLOATING = floating-leaved biomass in kg wet mass m-2
ACHS = ACHEX + ACHLIN + ACHLINCU + ACHMIN
S-NAVS = NAVGOT + NAVLAN + S-NAVPU + S-NAVRA
+NAVSUBTS
NITZS = NITZAM + NITZCAP + NITZFRUS.
The values for all taxonomic groups, whose acronyms are defined in
Appendices 1 and 2, are expressed as a percentage of the diatom
assemblages. This model appeared to have a significant relationship
with submerged biomass (R2 = 0.502, p = 0.034). Partial correlation
coefficients showed that floating-leaved biomass was significantly
correlated with the model (r = 0.882, p < 0.001, n = 22) when the








effect of submerged biomass was held constant. The model was not
significantly correlated with submerged biomass (r = 0.431, p =
0.051, n = 22), however, when the effect of floating-leaved biomass
was held constant. This indicates that the above model predicting
floating-leaved biomass is not confounded by the variable
submerged biomass. The correlation coefficient between submerged
biomass and the model, however, failed to be significant at the a =

0.05 level of significance by a marginal amount, and the model may
prove to be confounded in certain applications.


Submerged Biomass
Stepwise multiple regression was performed on percentage data
for 20 diatom taxonomic groups (Appendix 3.10) to predict
submerged macrophyte biomass. Cp statistic values indicated that
the best model contained 11 diatom taxon variables and explained
91.6% of the variance in submerged biomass (p < 0.001, n = 22). The
adjusted R2 for the model was 0.824. This model, however, also
explained 86.9% of the variance in floating-leaved biomass (p =
0.004, n = 22) and seems to be confounded by that variable. The
partial correlation coefficient between the model and submerged
biomass was significant (r = 0.960, p < 0.001) when the effect of
floating-leaved biomass was held constant. The partial correlation
coefficient between the model and floating-leaved biomass (r =
0.938, p < 0.001) was also significant when the effect of submerged
biomass was held constant. The model to predict submerged
biomass, therefore, is not a useful predictive model because it is
significantly confounded by floating-leaved biomass.








Sum of Floating and Submerged Biomass
Because submerged macrophyte biomass could not be separated
from floating-leaved biomass in predictive models, an attempt was
made to combine these macrophyte variables in a single predictive
model. Sixteen diatom taxonomic groups (Appendix 3.11) were
selected that appeared to show response to both of these macrophyte
variables. The stepwise multiple regression procedure was applied
to these groups and it produced the following best model:

FLOAT-SUB = 13.292 0.384(S-ACH) 1.159(S-AULA) -
23.126(EUNPEC) + 0.921(S-FRUSRH) 0.912(S-
NAVS) + 7.115(S-STAU) +2.492(EUNVAN) 3.6
R2 = 0.592, p = 0.042, n = 22
in which FLOAT-SUB = sum of floating-leaved and submerged
macrophyte biomass in kg wet mass m-2. Acronyms for the
taxonomic groups are defined in Appendices 1 and 2. The adjusted
R2 for this model was 0.389.


Emergent Biomass
The stepwise multiple regression procedure was performed on
percentage data of 17 diatom taxonomic groups (Appendix 3.12) to
predict emergent biomass. Cp values of all models with significant R2
values were much less than the p values, which indicated that these
models were subject to substantial random error.


Results of Canonical Correspondence Analysis
The canonical correspondence analysis option of CANOCO (ter
Braak 1987) produced eigenvectors (Table 9) ordinating the 47









Table 9. Eigenvectors for percentage data of 47 diatom taxonomic
groups in CANOCO Axisl constrained by percent-volume
infestation. Eigenvectors are presented in descending
order. Taxonomic acronyms are defined in Appendix 2.


Taxonomic Eigenvector Taxonomic Eigenvector
group group


S-GOMA
S-STAU
S-NAVPU
FRAGPIN
S-FRAGCO
S-EUN
S-ANOM
S-NAVSEM
S-CYM
TABFEN
NITZFRUS
S-EP
S-ACH
CYCSTEL
TABFLOC
S-SYNRUM
S-NAVRA
NITZFONT
CYCMEN
NAVGOT
FRAGCROT
COCPLACL
S-FRUSRH
NAVLAN
S-NEI
CYCSTELO
S-PIN
NITZPAL


1.78
1.34
1.32
1.20
1.14
1.00
0.77
0.54
0.50
0.45
0.43
0.31
0.10
0.06
-0.02
-0.02
-0.06
-0.08
-0.10
-0.10
-0.11
-0.21
-0.21
-0.23
-0.26
-0.27
-0.32
-0.35


AULAAM
ACTPUNC
NAVCUS
NITZAM
CYCPSEUD
NITZCAP
S-AULAGR
AULAISL
NAVSUBT
S-SUR
AULAITAL
SYNDEL
AULADIS
NAVCONF
CYSTEPDU
CALSPA
SYNFILEX
S-AST
FRAGBREV


-0.40
-0.44
-0.45
-0.52
-0.60
-0.61
-0.70
-0.76
-0.76
-0.81
-0.85
-0.90
-0.96
-0.96
-0.99
-1.04
-1.10
-1.55
-1.55








diatom taxonomic groups into an axis constrained by percent-volume
infestation. When axis scores for the 29 survey lakes were regressed
with percent-volume infestation values using the SAS GLM
procedure (SAS Institute, Inc. 1985), the following predictive
equation was obtained:
percent-volume infestation = 33.8+ 0.7(Axis 1) 3.7
r2 = 0.600, p < 0.001, n = 29
where Axis 1 is the sum of products of eigenvectors and percentages
of the 47 diatom taxonomic groups.
When canonical correspondence was invoked to produce
eigenvectors ordinating the 47 diatom taxonomic groups into an axis
constrained by percent-area coverage, CANOCO returned eigenvectors
ordinating 45 of the taxonomic groups (Table 10). Tabellaria
fenestrata and T. floccolosa may have been eliminated by CANOCO in
this ordination because of under-representation. Axis scores for the
22 survey lakes with percent-area coverage values were regressed
with percent-area coverage using the SAS GLM procedure (SAS
Institute, Inc. 1985). The resulting equation was:
percent-area coverage = 53.6 0.5(Axis 1) 3.8
r2 = 0.447, p < 0.001, n = 22

where Axis 1 is the sum of products of the percentages and
eigenvectors of the 45 diatom taxonomic groups shown in Table 10.


A New Predictive Model for Water-Column Total P
Because inferences from Whitmore's (1989) TSI(TP) predictive
model cannot detransformed to yield total P values for WCP (Canfield
et al. 1983a) estimates, a new model is presented here using the









Table 10.


Eigenvectors for percentage data of 45 diatom taxonomic
groups in CANOCO Axisl constrained by percent-area
coverage. Eigenvectors are presented in descending
order. Taxonomic acronyms are defined in Appendix 2.


Taxonomic Eigenvector Taxonomic Eigenvector
group group


FRAGBREV
FRAGCROT
NAVCONF
S-NAVPU
CYCPSEUD
CYSTEPDU
NITZPAL
CALSPA
SYNFILEX
AULAAM
NITZFONT
S-AULAGR
AULAITAL
S-AST
COCPLACL
NITZCAP
NAVCUS
CYCMEN
NITZAM
AULAISL
S-NAVRA
NAVGOT
NAVLAN
S-SYNRUM
S-NAVSEM
CYCSTE
NITZFRUS
SYNDEL
AULADIS


2.11
1.49
1.47
1.26
1.17
1.16
1.15
1.06
1.01
1.00
0.88
0.86
0.86
0.85
0.84
0.59
0.55
0.53
0.53
0.43
0.42
0.25
0.21
0.19
0.18
-0.07
-0.10
-0.15
-0.16


S-NEI
S-ACH
S-PIN
S-STAU
S-FRAGCO
NAVSUBT
CYCSIELO
FRAGPIN
S-SUR
S-FRUSRH
S-ANOM
S-GOMA
S-EP
S-EUN
S-CYM
ACTPUNC


-0.17
-0.18
-0.19
-0.25
-0.28
-0.30
-0.31
-0.35
-0.48
-0.71
-0.73
-0.84
-0.86
-1.08
-1.18
-1.29














1.00-

0.75-

0.50-

0.25-

0.00-

-0.25 -


-0.50 -

-0.75 -


-1.00 -
-3.5


+
+ +



++
**



+ oo



o N-limited

* + balanced

P-limited
*


I I .
-3.0 -2.5
Log


Figure 9.


I I
-2.0 -1.5
limnetic


I -
-1.0 -0.5
total P


10
Log-transformed total P versus
log-transformed TROPH1 diatom
index for 51 lakes.


0.0








TROPH 1 diatom index (Whitmore 1989) to predict log-transformed
water-column total P. Figure 9 shows a plot of log-transformed
water-column total P versus the TROPH1 diatom index for 51 Florida
lakes. The four nitrogen-limited lakes shown demonstrated
anomalously low diatom index values for their total P values, and
were removed from the regression data set. The resulting predictive
model was:
logo0(total P) = -1.795 + 0.973(loglo(TROPH 1)) 3.9
r2 = 0.807, p < 0.001, n = 47.


Assessing Confoundedness in the Water-Column Total P Predictive
Model
The significant negative correlation coefficient that was observed
between log-transformed total P and percent-area coverage (Table
11) confirmed the inverse relationship previously observed (Canfield
et al. 1984) between macrophyte presence and water-column
nutrient concentrations. The correlation coefficient between the log-
transformed TROPH 1 diatom index and percent-area coverage,
however, was not significant (Table 11). The model for predicting
water-column total P from the TROPH 1 diatom index is not
confounded by the macrophyte variable percent-area coverage.








Table 11.


Correlation coefficients between TSI(TP), percent area
coverage and the log-transformed diatom index TROPH 1.
* = p < 0.05.


correlation coefficient
prob. > 1 r l under Ho: rho = 0
sample size


Water-column
total
P


log(TROPH 1)



Percent-
area
coverage


*0.838
<0.001
29

*-0.537
0.010
22


Percent-
area
coverage


-0.359
0.101
22













CHAPTER 4


DISCUSSION



Dominant Environmental Variables and Scale of Analysis
The purpose of this study was to identify diatom taxa that are
indicators of macrophyte presence in lakes. Despite an appropriate
sampling design, macrophytes were not found to be a primary
determinant of diatom assemblages. A trophic-state/pH/specific-
conductance environmental gradient was the principal influence on
the diatom communities.
I sampled lakes along a wide gradient of macrophyte abundance
in order to develop predictive models with broad application for
inferring historical macrophyte standing crop, and by necessity I
transgressed the trophic-state gradient that is present among Florida
lakes. A negative correlation exists between macrophyte abundance
and water-column Chl a (Canfield et al. 1984) in Florida lakes. I also
observed a negative correlation between percent-area coverage of
macrophytes and water-column total P in the present study. Lakes
that were low in percent-area coverage (< 10%) were high in water-
column nutrients, whereas several lakes that were high in percent-
area coverage (>80%) were low in water-column nutrients.
pH, trophic state and specific conductance were mutually
correlated in this study as in other studies of Florida lakes (Canfield








1981, Brenner et al. 1990, Whitmore 1989). The majority of Florida
lakes are soft-water, acidic and low in alkalinity (Brenner et al.
1990). Nevertheless, many lakes on phosphatic sands or carbonate-
rich bedrock are naturally high in productivity and dissolved solutes
(Canfield 1981). Because the present survey spanned a wide range
of limnological conditions, trophic state, pH and specific conductance
formed a dominant environmental gradient that emerged as the
principal determinant of diatom community composition.
Jackson and Charles (1988) observed the effect of a similar
environmental gradient on macrophyte distribution in the
Adirondack lakes of New York. They found that alkalinity, pH and
ionic composition were interrelated factors that determined the
distribution and species composition of aquatic vegetation. Jackson
and Charles stated:

We conclude that the chemical gradient underlying
compositional variation among our Adirondack softwater
sites is the tail end of a broad pH complex-gradient that
extends to highly alkaline waters. At the scale of
environmental variation observed in Adirondack lakes,
the main factors associated with vegetation variation are
pH, alkalinity, Ca, Mg, and perhaps Al. In regions where
the gradient is broader, or at least where the hardwater
portion is represented, conductivity and trophic status
become more important (Jackson and Charles 1988, p.
1456-1457).
Because Florida lakes exhibit a wide scale of variation from
softwater to hardwater conditions, the resulting chemical gradient
determined diatom community composition in the same manner that
the chemical gradient in Adirondacks lakes determined macrophyte
composition. The fact that diatom community composition was








determined mostly by trophic-state, pH, and specific conductance
rather than by macrophyte presence was thus the result of the scale
of analysis as discussed by Duarte and Kalff (1990). The dominant
effects of trophic state and pH may have overridden differences in
community structure that resulted from the influence of
macrophytes. In addition, most lakes in the survey were shallow
(mean depth < 3.0 m), and periphytic diatom communities may be
less specific about substrate types than previous qualitative studies
(e.g. Round 1956) have suggested.
Future studies might minimize error variance in macrophyte
predictive models by focusing on a calibration set of lakes that
covers a narrower range of trophic state and pH. This approach
would sacrifice generality but improve the precision of prediction for
lakes with a limited range of macrophyte standing crop. Duarte and
Kalff warn, however, that "There is no reason to expect that patterns
found at any one scale are transferable to other scales" (Duarte and
Kalff 1990, p. 362). A problem of scale arises when any predictive
model derived from a set of limnologically diverse lakes is applied
historically to a single lake that has remained comparatively constant
in character over time. Confidence intervals are inappropriately
large for historical predictions because fewer factors affect the error
variance within a single basin than within a calibration data set.


Negative Relationship Between Chl. a and Macrophytes
Lower water-column total P values may occur in lakes with high
macrophyte standing crop for the following possible reasons invoked
by Canfield et al. (1984) to explain macrophytic influence on Chl a:









1) The phytoplankton community competes with macrophytes,
especially floating macrophytes (e.g. Eichhornia), and their
associated epiphyton for dissolved nutrients in the water
column. High macrophyte standing crop could therefore depress
water-column nutrient concentrations;
2) Macrophytes minimize wind mixing and resuspension of
nutrients from bottom sediments leading to a reduction in
nutrient cycling.

Other possible explanations for the negative correlation include:

1) Rooted macrophytes may proliferate in lakes that are naturally
low in water-column total P concentrations because they don't
depend on the water for their nutrient supply. Most of the P
they utilize is derived from the sediments (Carignan and Kalff
1980);
2) When water-column nutrient concentrations are high,
phytoplankton and epiphyton standing crop increases and may
limit submerged macrophytes by shading (Sand-Jensen and
Sondergaard 1981).


Response of Periphvton to Water-Column Nutrients
TSI variables demonstrated positive correlations with the
proportion of planktonic diatoms, and negative correlations with the
proportion of periphytic diatoms in the survey lakes. This suggests
that planktonic diatom populations assume greater importance
relative to periphyton populations in lakes that are higher in trophic
state. The positive correlations between water-column total P and
the concentrations and log-transformed accumulation rates of
periphytic diatoms, however, show that even if plankton assume
greater importance at higher water-column nutrient concentrations,
periphyton production also increases with increasing trophic state.