"

,,,, <\~:ti ;'

.;:~ ":

',

,, . ~t ' ·' '. ·'

~ ·-: ; -,

-:" ·r

~ ..

-:A-:·

:~.~~( f li ,, " '! ~

•o,,

'·

'' (

' l

h\ ·~~ ,;,.:, ~J';·

F t • •• ~i

.... ~~~-~:"' "t \ \;

lf' ,f;_. -

tt ... ,.

': ,, -· '

-~·

.~! "lt~t' -'-

. .,_._

'- _.,

.. IDENTIFICATION OF MUSKRAT CQND~~RA ~IB§!HI2Q~) HABITAT IN RIVERINE

ENVIRONMENTS1

ROBERT P. BROOKS, 2 Department of Forestry and Wildlife Management, versity of Massachusetts, A~illerst, Massachusetts 01003

Uni-

wENDELL E. DODGE, Massachusetts Cooperative Wildlife Research Unit, Uni­versity of Massachusetts, Amhe::.•st, Massachusetts 01003

ABST~AC!: Muskrat (Qnda_!_£~ !ibe_!!:!.!~~!) have been studied extensively, --bUt with little emphasis on riparian populations in rivers. Watersheds

harbor a substantial portion of the buz-row-dwelling population; there­

fore, identification of suitable habitat would assist in estimating abun­

ctance. A list of ecological factors of potential importance to muskrats

was obtained from information in the literature and watersheds that pro··

vided a diversity of riverine environments were chosen for study. The

number of active burrows occurring in 74 randomly selected sites, each

aoo m long, was determined for 5 rivers in Massachusetts during the

sUIIJllers of 197 8 - 197 9. River hydro 1 ogy, vegetative cover, bank st rue­

tore, and landuse components were measured quantitative.ly for each site.

Key habitat factors were determined by principal components regression.

A linear model was developed to predict potential muskrat habitat. The

numbers of islands and coves were positively related to muskrat abun­

dance. Riv~r banks bordered by open and agricultural land were selected

more often than forested banks. High water velocity and poor bank sub­

strate were negative influences. Bank height and slope were unimportant

above required minimum values of 0.2 m and 10°, respectively. The appli­

cability of using rem?te-sensing sources to delineate these 1<ey factors

was tested on 3 additional rivers. Landuse and topographic map~ were

used to select favorable habitat. In every instance, more burrows were

found at sites chosen according to important factors from the predictive

-:Jodel, than at sites selected at random. The relationship of frequency

'

1Research cooperatively supported by Massachusetts Cooperative Wildlife Research Unit (U.S. Fish and Wildlife Service, Massachusetts Division of Fisheries and Wildlife, University of Massachusetts, Amherst, and Wild­life Management Institute) and a University of Massachusetts Fellowship.

1

"Present address: The Pennsylvania State University, The DuBois Campus, ~liege Place, DuBois, Pennsylvania 15856.

113

·.J :'j

,.

/j

:t.··· \' ., ,, _j·:, •' ,.' • ' <

,&

: '> ' _,..,.- - ' ' ' . '~l'!>i:·; •q;. ~'-~ ... ' . .. . . . ' ' ' ' ' ' I ~~-~-...::~ ·. .

-,;

!/ --.'~ /i

.--··· ~.A

I i I

J

I l l

I u

I I

1

BROOKS AND DODGE

and duration of river flooding to muskrat habitat selection was investi~

gated by comparing discharge with bank height, and seasonal discharge

with statewide harvest data. The a.ppl ication of the model in estimating·

distribution and abundance of muskrat habitat in a watershed is dis~

cussed.

-------------------------------------------------------------------------

The muskrat CQ!_!da_!!~ !.~be_!~_!~~~) is, both in numbers and income~ the most

important furbearer in the world (Willner et al. 1975). Habitat for this

cosmopolitan species includes both lentic and !otic waters. Muskrats

build 2 types of permanent shelt~r: a burrow or den dug into banks of

lakes, ponds, or rivers, and a house or hut constructed of herbaceous

plants and mud in shallow, lentic waters.

Research has focused on marsh and pond populations; however, riparian

muskrats form a substantial portion of the total muskrat population, but

they are difficult to observe and census.

Although direct census methods of estimating animal populations are pre­

ferred, these methods are not feasible for use with most small, covert

mammalian species, such as muskrat. Therefore, in the 1st phase of the

study (Brooks 1980) reported here, muskrat distribution was studied as a

function of requisite habitat features that could be detected from

remote-sensing sources. Thus, the major objective was to identify key

physical and biological characteristics that determine habitat selection

by burrow-dwelling muskrats in riverine environments.

MATERIALS AND METHODS

Study Area

A discrete watershed (i.e., where the r!vt~r's mainstream is about a 4th

or 5th order stream; Leet and Judson 197lj was chosen to represent an

area of eco 1 ogi cal importance to muskrats, and pro vi de an opera t i anal

unit for biologists. Eight watersheds that constituted a variety of

riverine environments were studied in Massachusetts (Table 1). Rivers of

114

' . . . . . ... ~ • J

··~

J

o$

Cll .... +' Q.) Cll ~

..c:: <::)

o$ Cll

j Cll

"' ~ l c l •.-4 J

.... l o$ +' •r-4 ! .a o$ l ..c: +' l o$ $-1

I ~ Cll :;j a

I ~ 0

'+-1

"0 l . Q.) r

.I - I ~ l

I

o$ l Cll

I Cll ~ Q.) I > I •.-4 $-1 !

) I

'+-1 I 0

Cll j <::) l •.-4 .... l Cll ...... i $-1 i Q.) ! +' c.~ I o$ t $-1 o$

..c: <::) -o$ <::) ...... en ~

..c: ~

.-i

Q.) --.0 a:l

E-4

I • ., • • . ' ~· • . • ~ ~ • • ' . , . •

~•"!itl!~ ·r.r, f ll§#!frP~21i!J ~! _(J,u,

~ •. , 'l i

~----·--· _ _M

8.5 8

\

- -- ---- ,.----;-::-;-.

.lib =q - ·fi!Yi!lliii

Table 1. Physical characteristics of rivers sampled for muskrat habitat in Massachusetts.a

River

Name

----------------------------------------------------------------------------Total

Drainage

Area (km2 )b

Mainstem

Length (km)

Gradient

(m/km)

Mean

Discharge

m3 /s(cfs)

Number of

Sites

Sampled

------------------------------------------------------------------------------------------------

Chicopee 1865 33 2.4 25.5(899) 4 Ware 567 55 2.1 8.2(289) 50 Quaboag 544 60 3.3 6.9(242) 8

Nashua 1375 67 0.6 15.9(561) 8 Manhan 148 39 4.7 - 4 . Concord 1049 25 0.8 17.6(620) 2

Ass abet 453 51 1.2 5.2(183) 6 Housatonic 1295 111 1.4 15.0(528) 8 (in Massachusetts)

t;J ~ !;;::

:::: ~ '-! ~ ~ a !;;::

a ~

~

§ ~ ~ J..:!

~ ~

-----------------------------------------------------------------------·------------------------ ~ ~

Mean or (total) 912 55 2.1 13.5(475) (90)

----------------------------· .. - ------------------------------------------------------------------aData from USGS topographic quadrangles and USGS (1979).

bRepresents entire watershed, including some major tributaries not studied.

.-::_'" .i:;.,..,.:;~·v,..~Jitji~,;~ ~v~c"·:•· ·

J

!L· < '\'-'=

,

~· (

\\

~ .

I ~(~t-;ttl!fi

I

~.i

c

}}

BROOKS AND DODGE

r .)oth gradual and steep gradient were represented. Fifty study sites were

randomly selected on the main study area, the Ware Rive~·. in summer 197a I

and 24 additional sites were selected on 4 other rivers in summer 197s.

From data collected on these 5 rivers, a habitat model was developed,

This model was tested on 16 sites on 3 additional rivers (i.e., Concord I

Assabet, Housatonic) in September 1979 (Table 1).

Habitat Analysis

Activity and home range estimates from the literature (Errington 1940 I

Mallach 1971, Stewart and Bider 1974) suggested that sections of river

300m long would be appropriate as study sites. Both shorelines of each

study site were carefully searched for muskrat burrows by walking slowly

through the water and probing above and below water level with a hardwood

pole. River bends, coves: and islands increased the amount of shoreline

checked. Muskrat signs were important for determining burrow activity

a.nd interpreting use of the habitat. Active burrows provided the best

permanent indicator

signs and activity.

proximity, about 2 -

of muskra presence and were identified by recent

Each hole was considered a burrow, and holes in

3m maximum (Earhart 1969), were considered to be an

interconnected burrow system. In the present study, reference to a bur-

row means burrow system unless otherwise noted.

Information on hydrology, soils, vegetation, landuse, and muskrat signs

was collected for each site as a function of burrow abundance. A 2nd set

of more specific data was collected for the microenvironment of each bur­

row within each site for comparison with general site characteristics.

Methodology for collecting data on sites and burrows is presented by ex­

ample (Figure 1- 2). Measurement techniques were described in detail by

Brooks (1980).

Data collected from sites were used to select 16 habitat variables that

appeared important to muskrats or might be detectable from aerial photo­

graphs or existing maps. VR.riables that either had low variability or

duplicated other variables were deleted a E.!:!~E..! and were not used in the

' habitat model. Principal components regression (PCR) offered an al terna­

tive to ordinary least squares regression for constructing a model with

116

Figure 1.

. ' ' • ,. ' ' J~ • • • I f L ft

' . ' • ' • ' • • t 1 • "'

• - ... , • "" • - '\t" . . t . ., I ~ -, . . • •

EV.

,i~ --~.

I

j j r I J ,'!

l

I

\

:'l-

IDENTIFICA11ION OF MUSKRAT HABITAT

SITE EVALUATION

/

POOL

O•

"

I

SOIL SAMPLES

1m HI

I

I I

I I I

I

I I

I I I

,'TRAIL I I OPEN

meters

0 20 40

Figure 1. Example of riverine site inhabited by muskrats.

117

N

~

60

·~ I u·

SHADE

~-----------.------~ COVER

1- DEC ll)._~_SHRUB+-- OPEN --1 TREE:

0

WATER

HEIGHT

l

2 0 METERS

FINE

BURROW

EVALUATION

x VELOCITY

2

Figure 2. Example of river bank inhabited by muskrats.

118

1 eve 1:

a J. • 1L

sHAZAI'

signi.t

water~

Housa11

on u.e series;

racy a

W'1S ca

dures

for raJ

To det

tr ibutf

River

along

cord~

Flood in

The eff

each si

tions

Massach1

percent1

were de

dischar& muskrat

seasonal

daily di

In the . years, 1

foi." Mass

I

lc

)

(;

' •)

,:_)

I( {t

.••• ~.:~~::_~~__.....;..".;· ... ,.;~--.--· ..... ~;o'JIII·~·'l1 ;1'\' "~~~ ... ---=··~-·· -~· . J

j

J

I ·r ·' l

I I . . ! ~

l

IDENTIFICATION OF MUSKRAT HABITAT

interrelated habitat variables (Willis et al. 1978), and significance

levels could be determined accurately (Freund and Debertin 1975, Hill et

al· 1977). The statistical packages BMDP (Dixon and Brown 1979) and

sHAZAM (White 1978) were used for these analyses.

significant variables (P < 0.05) taken from the PCR habitat model on 5

watersherls were used to predict favorable habitat for muskrats on the

Housatonic "''.d Concord-Assabet river systems based on features depicted

on p.S. Gevlo~ical Survey (USGS 1979) topographic quadrangles (7.5 min

series) coverin~ the rivers sampled. In addition, to determine the accu­

racy of the habitat model, the number of burrows predicted foi' each site

was compared with the actual number of burrows observed.

dures were used to verify the habitat model with respect to

for ranking suitability of muskrat habitat.

These proce­

its efficacy

To determine minimum stream size requirements for muskrats, we checked

tributaries for signs of muskrats from their confluence with the Ware

River to 1,000 m upstream. Measures of gradient (i.e., drop in m/km)

along the 1,000 m length and discharge at the stream's mouth were re­

corded.

Flooding

The effects of flooding on muskrat abundance could not be measured for

each site, but because flooding increases mortality in n;,}~kra t popula­

tions (Errington 1937), we estimated the significance of flooding on

Massachusetts rivers by 2 methods. Stag-e duration curves, indicating the

percentage of time that water level remains at a particular gage height,

were developed for 3 USGS gaging stations on the Ware River. A c_ritical

discharge value (i.e., one that nearly crests a river bank, thus flooding

mskrat burrows) was obtained from the curves. Frequency, duration and

seasonality of critical discharges for the Ware River were obtained from

daily discharge records kept by USGS.

In the 2nd approach, we compared statewide harvests of muskrats for 13

years, 1965 - 1978 (Chet McO<>:rd, pers. CQmm.), with an index of discharge

for Massachusetts based on 8 majotr watersheds. A PCR model baf}ed on mean

'·

•·

I

I

r- 11

BROOKS fiND DODGE

seasonal discharge was developed. 5ignificant discharges were related to

ctitical times of the year for muskrats.

RESULTS

Six variables were significant (P< 0.05) in the final habitat model, but

only 5 could be detected on USGS quadrangles or other remote-sensing

sources. Small changes in bank slope were undetectable, so this charae~ teristic was not used as a predictive variable. MacConnell and Nied~ zwiedz (1979) found that classifying bank height and slope by 3 - m in~ tervals was difficult at 1:12,000 scale. Substantial ground reconnais~

sance was required for accurate measurements.

The final habitat model is presented as Equation 1:

Y = -0.726 + 0.568 (ISLANDS) + 0.813 (COVES) + 0.008 (OPEN) p

- 0.005 (FOREST) + 0.531 (CLASSDOM)

where,

(1)

Yp is the predicted number of active muskrat burrow systems per

site, ISLANDS is the number of islands per site,

COVES is the number of coves per site, OPEN is the percentage of open and agricultural land per site,

FOREST is the percentage of forest land per site, and

CLASSDOM is the dominant river class per site.

An ave1•age Yp' (Yp), per site for a watershed was calculated by using all

sites analyzed in that watershed (Table 2).

To test the habitat model, we used 4 variables that were positively

related to burrow abundance (i.e., ISLANDS, COVES, OPEN, CLASSDOM) to se­

lect favorable muskrat habitat on 3 additional rivers. The total of 29

burrows observed in 8 favorable sites was significantly greater than

(P<0.01) the 9 burrows observed in 8 random sites on the same rivers.

120

'.- .. . . . •' . . .q ' ,I

' - . '- . . '

ware

Manha;

Nashu.

con co·

Quabol

Chi CO}

-----Mean .. (not ~

indic

burro

Evalu

and d

tat.

charg•

bot tor

.:•n s t

m3/s),

sub sty

gradie

respec

' i

I

I I

I i

I

• t I J

IDENTIFICATION OF MUSKRAT HABITAT

Table 2. Performance of habitat model for predicting riverine muskrat burrows on a watershed basis based on adjustment.

------------------------------------------------------------------------Mean number of burrows per site

Observed Predicted (Equation 1)

Adjusted (Equation 2)

~---------------------------------------------------------------------

Housatonic 2.88 0.83 2.88

Assabet 2.17 0.82 2.74

ware 1. 92 0.75 1. 92

Manhan 1.75 0.73 J. 74

Nashua 1. 25 0.66 1.25

concord 1.00 0.48 0.74

Quaboag 0.75 0.51 0.78

Chicopee 0,50 0.29 0.50

--------------------------------------------------------------------~----

Mean + S.D. '(not weighted)

1.53 + 0.80 0.63 + 0.19 1.57 + 0.91

------------------------------------------------------·------------~------

indicating that the PCR model identified important habitat factors for

burrow-dwelling muskrats from remote-sensing data .

Evaluation of 19 of 38 Ware River tributaries indicated that g:t"Jidient

and discharge determined the importance of small streams as muskrat habi­

tat. Muskrats were present when gradient was low ( < 6.1 m/km) and dis­

charge exceeded 0.11 m3 /s (4.0 cfs). These streams had fine to organic

bottom substrates, and fN(~'J!e dominated by wetlands. Muskrats were absent

on streams with high gradient ( > 9.0 m/km) and low discharge ( < 0.11

m3/s). These streams were characterized by rocky, coarse, or fine bottom

substrates and occurred in forested and urban areas. For streams with

gradients of 6 - 9 m/km, discharge should be the deciding factor with

respect to muskrat habitation.

121

I

,,

It) a:; co

I

!' j

1 i I

I I

. ''

BROOKS AND DODGE

Flooding

Flooding was hypothesized to affect muskrat survival and therefore have

an effect on selectiQn of habitat. The gage height (stage) corresponding

to bank height in each section of river was determined. Variable ban}{

height along the Ware River and relatively steep banks, however, pr~~

vented an accurate determination with respect to gage height. Rivers

having wide floodplains (i.e., older river valieys:, unlike the Ware

River, would tend to have a more regular bank height caused by formation

of natural levees. Increases in stage measurements would be less notice~

able as flood 1 evel s exceeded bank height and waters flowed onto the

floodplain. At this level of discharge, a stage duration curve would

show an inflection point. No obvious reductions in the rate of stage

increase were observed on curves for the 3 Ware gages. Discharge co 1·~

responding to flooding over the banks occurred about 20 percent of the

time for a11 3 gages (Higgins 1967). These discharges occurred most

commonly in either March or Apri 1, or December - Apri 1 from 1965 - 1978,

The PCR model of the 4 seasons and harvest data indicated that both

winter discharge of the year preceding the harvest and autumn discharge

in the year of harvest were positively significant (P < 0.01, R2 = 0.59).

High discharges from December - February and September - November may

signal a forthcoming year of high muskrat harvest. The spring population

remaining may respond with an increase in reproduction to compensate for

high winter mortality (Errington 1951).

The effects of flooding should be studied further with respect to sea­

sonal muskrat survival and locations of burrows which can be flooded for

variable periods of time, particularly where banks are low.

DISCUSSION

The initial phase of this riverine muskrat study was to identify key

cha ... •acteristics of suitable habitat that might be us.eful for predicting

populations on a watershed basis. The final habitat model (E;quation 1)

predicted favorable muskrat habitat and, to some degre•e. the abun.cllince of

muskrats.

122

y---· ·lJ nr The

tnan E

able 1

10 eitY

coves

cludin

The ob

can a bot torr.

itself

1entic

be enc

rats <

tincti

Island

burro\\

such

pres en

but th

The de

select

avoi

numb.

exam in

water

rats,

condit

as fot

muskra

slower

by mus

and .r:

ship

finer

stream i

I

~co )

IDENTIFICATION OF MUSKRAT HABITAT

The number of coves (COVES) was more strongly associated with burrows

than anY ath~r variabl~. Besides inoreasing the length of shore avail­

ab1e for burrows, coves provide shelter from river currents of high ve­

locitY and a readily avai !able food source of emergent vegetation.

coves can be identified from a variety of remote-sensing sources, in­

cluding USGS topographic quadrangles and panchromatic aerial photographs.

The obvious signature of water and aquatic Vegetation on infrared imagery

can also be used to identify coves. Coves, where water conditions,

bottom substrate, and vegetation usually differ from those in the river

itself, can be envisioned as an appendage to the main river channel. If

Ientic conditions and abundant emergents prevail, muskrat houses may also

be encountered. A cove differs from a land-locked marsh because musk-. rats can travel freely by water between the cove and river. This dis-

tinction may explain periodic population shifts up or down stream.

Islands (ISLANDS) also increased the length of shoreline available for

ourrowing. Although islands can be flanked by unfavorable conditions,

such as rapids, more often a section of river of slower velocity is

present. Neither coves nor islands were comnon on the rivers studied,

but th•ose present served as foci for muskrat burrows.

The dominant class of a river site (CLASSDOM) strong·ly affected habitat

selection by muskrats. Rapids, rated zero in the habitat model, were

avoided by muskrats, and therefore contributed nothing to the tctal

number of burrows, per site. Rapids dominated 11 percent of the sites

examined, but only 1 burrow located in an adjacent cove was found. As

water velocity slows a section of river becomes better suited for musk­

rats, until at a rating of 3, bnckwater areas or coves provide the best

condition3. Backwater areas dominated 4 percent of the sites but served

as foci fa.r 20 percent of the burrows, illustrating their importance to

muskrats. River class reflects a preference by muskrats for sections of

slower velo.ci ty ( < 10 m/min). Other studies have indicated a preference

by muskrats for pools and backw~ter areas (Errington 1937, Warwick 1940),

and rivers of slow velocity (Crawford 1950, Byrd 1951). This relation­

ship results in burrows be]ng located where bottom substrates are of

finer material or vegetated. Zejda (1976) found that muskrats avoided I

streams with coarse substrates.

r l

,

.r' . . . . 1 I ~~~:_J __ =-~---- -, i BROOKS AND DODGE ·~·---~-~---, --~~"1 I

, ·{ . River·/! i Landuse affects abundance of muskrats. The predominantly forested land-! 110 t s I· j scape of Massachusetts explains the high percentage of forest land found

',1 · + h k h h d · · f · t 1 me 11 t s 1 along rivers. Fores"ed _an S 1 owever, a a s1gn1 tcan y negativ ' ! . e gra __ die j effect on burrow location (FOREST, Equation 1). Al tnough forests donu-

sra f nated overall, muskrats selected banks with herbaceous cover for 40 Per-1 from U 1~ cent of their burrows. The percentage of open land had a positive in-

f f charge J fl~ence on burrow location (OPEN, Equation 1). A pre erence or emergent

vegetation by muskrats is well-known (Willner et al. 1975), but herba-

ceous plants compose a substantial part of their diet (Errington 1941),

particularly along rivers. This preference for open land agrees with

research from other regions (e.g., Errington 1937, Gilfillan 1947, Byrd

1951, Zejda 1976). Due to this preference, 34 percent of all burrows

studied (n = 160) were partly supported by roots of herbaceous plants •

The habitat model identified favorable muskrat habitat, but other habitat

variables were also important. Bank structure and composition affected

locations of burro'IS. No burrows were found where bank height was less

than 0.2 m and/or bank slope was less than 10°. Apparently, it is phy­

sically difficult for muskrats to construct and maintain a burrow syst~

under these conditions. Earhart (1969) found a 10° limit in slope, but

claimed 70 percent sand was also required. Mean percent sand exceeded 70

percent for both sites and burrows in this study. Heavily vegetated

banks along New England rivers and the resultaat organic matter in the

soil probably bind sand grains together better than in California where

Earhart studied burrows in soils with little organic material. Bank

soils composed of high percentages (90- 100 percent) of sand and gravel

would be unaui table for muskrats in New England. Therefore, banks are

suitable for burrowing, provided height and slope are sufficient and per~

centage of sand or gravel is less than 90 - 100. River banks that are

riprapped or covered with impervious material, as in urban areas, prevent

construction of burrows and therefore should be considered void of bur­

rows in a survey of such areas. Banks of low height and slope usually

occur in conjunction with organic soils in wetlands adjacent to the

river. These sections may be suitable for houses, but provide poor sub­

strate for burrows. Thus, the variable for wetlands (GNDWET) did not

relate positively to burrow abundance.

124

addres I maY bE i

are PJ j

t ions .1 j

erosic l l

reache I

ent,

slow 1

annual

muskrE

The f

riverf (Tabl

per v

Housa

under

cient

basis

t ion

1 to

predi

ship

t ion

By .st tat n t ion

of bt

obser

. t;

t

L .... 1.

I ; j '\ . l

··1iiiiliilll'lll.,l

I

i c

~-· ~·~-----~---------.. -.:...---·--·-------------~---··--·-·--------.....:._

IDENPIFICAPION OF MUSKRAP HABIPAP

River width, used to assess river habitat in Ohio (Gilfillan 1947), was

not significant in the final habitat model. Minimum stream size require­

ments for tributar-ies3

to the ~iare River for muskrats were < 6.1 tn/km

ctient and >O.j_1 m Is discharge. These streams were often 1st and gra

3rd order streams with some associated wetlands. They can be selected

from USGS topographic quadrangles (7.5 min series) if an estimate of dis­

charge is available. Although the maximum river width that should be

addressed by the habitat model is not certain, the following suggestions

Y be useful. Rivers as large as the Connecticut or Merrimack proper ma

are probably poor habitat for muskrats due to sudden, violent fluctua-

tions in depth from regule.~ion of hydroelectric dams and scou1•ing and

erosion phenomena along the banks. The Chicopee River proper and lower

reaches of the Concord and Nashua Rivers, where few muskrats were pres­

ent, had similar characteristics on a smaller scale. Unless a river h.

sloW flowing and has excellent muskret habitat, streams with a mean

annual discharge >30 m3 /s (about 1,000 cfs.) are probably poor habitat for

muskrats.

The final habitat model (Equation 1) was reapplied to data from all

rivers to compare the numbers of burrows predicted with actual counts

(Table 2). The model underestimated burrows by an average of 52 percent

per watershed, more for rivers with prime habitat for muskrats (e.g.,

Housatonic) and less for poor rivers (e.g., Chicopee) (Table 2). This

underestimation was due to deletion from Equation 1 of variable coeffi­

cients unequal to zero, and absence of variables llvt measurable on a site

basis (e .. g., flooding, natural mortality factors). A mathematical func­

tion was developed to relate the number of burrows predicted by Equation

1 to the number of burrows observed during site i:1spections. Plotting

predicted values against observed values produced a curvilinear relation­

ship that was approximated by a 3rd degree polynomial expression (E~ua­

tion 2):

- -3 2 YA:: 27.45 Yp- 34.84 Yp + 15.60 Yp - 1.76 (2)

By substituting the mean predicted number of bu.rrows (Y ) from the habi-p

tat model (Equation 1) into the equation for correcting the undel•estima-

tion of burrow numbers (Equation 2), one obtains a mean adjusted numbe 1•

of burrows per site (YA) that corresponds closely to numbers of burrows observed (Table 2) ,

125

. . . . \ . . . . .. .. 'I '

~ ' • 1 / • l ' . . ... ... .

~

··1·'.' ~ ~

~ ,,

'l I

BROOKS AND DODGE

The adjusted habitat model can be reliaoly used to !'ank .civers With

respect to favorable habitat for muskrats and provide an index to poten,

tial abundance.

CONCLUSIONS

A model of habitat selection by burrow-dwelling muskrats was developed

for rivers in Massachusetts. The model was used successfully to predict

favorable habitat and provide an index to muskrat abundance based on top,

agraphic maps and aerial photographs. The number of predicted burrows

must be adjusted to accurately reflect burrow density on a water8hed

basis. Physiognomy of a river and its banks determine selection of habi,

tat by muskrats. Muskrats select sites with slow water velocity, suit­

able banks for burrowing, and sources of herbaceous and emergent vegeta­

tion. Backwater coves are particularly important centers of activity

relative to their availability. Banks must exceed 0.2 m in height and

10° in slope to allow burrow construction.

Muskrats are more abundant in rivers of intermediate width and discharge

than in either small or large rivers. They were not found on streams of

steep gradient (>9.0m/km) or low discharge (<0.11m3/s). Water level

fluctuations and bank erosion of large rivers (> 30.0 m3

/s) are not con-

ducive to colonization by muskrats.

The habitat model should be applicable to most temperate reg·ions where

muskrats occur because habitat variables used were not specific to Mass~ chusetts watersheds, and their importance for muskrats was supported by

the international 1 it era ture surv1ayed. We recommend that the mode 1 not

be applied in northern coniferous forests, riparian habitats in arid en­

vironments, or tropical climates until further field studies are com­

pleted. The apparent lack of muskrats in rivers of high turbidity in the

southeastern U.S. observed by Errington (1963) suggests that the influ­

ence of turbidity should be incorporated into the model in th1s region.

Suitability for muskrat burrows of soils that differ greatly from those

in the glaciated northeastern U.S. should be considered before the model

is applied.

126

~ . ' ' . • • ' . '• ' • ~ • • I

. . - .

Ar;pr.o,·

remotE

waterE

uiatjc

waters

ACKNO"

Apprec

work a

gestio

LITERA

BROOK

\

CRAWF< \'

DIXON. t

EARHAF c I

ERR INC

J

l l r ! l ' ! I l

' l

1 1 L

r ,

----a

p

----p FREUND

t 5'

GILFILJ t

HIGGIN: pt 54.

HILL, fl(

AI

l

I \

J

.I

rd.~~~·· ............ c· ..... .,..,~, . .,...,....,..~ .• te !-..:. t'1 ~___.,..._..~~.~~-J .. IDENTIFICATION OF MUSKRAT HABITAT

Approximately 20 percent of a watershed's length should be analyzed from remote-sensing data on a site basis to insure adequate comparison between

t ersheds. Marshes or ponds containing substantial house-dwelling pop­wa uiations adjacent to a river should also be considered when ranking

watersheds with respect to favorable muskrat habitat.

ACKNOWLEDGMENTS

Appreciation is extended to Lars Bors-Koefoed for assisting during field work and to Donald R. Progulske and Frederick Greeley for editorial sug-

gestions.

LITERATURE CITED

BROOKSs R.P. 1980. A model of habitat selection and population estima­tion for muskrats (Ondatra zibethicus) in riverine envirorunents in Massachusetts. Ph.D~The~s-.--Univ~Iassachusetts, Amherst, Massa­chusetts. xii + 113 pp.

BYRD. M.A. 1951. The economic importance of the muskrat in Virginia with particular emphasis on Montgomery, a mountainous county. M.S. Thesis. Virginia Polytech. Inst., Blacksburg, Virginia. 22-1 pp.

CRAWFORD, B.T. 1950. Some specific relationships between soils and wildlife.· J. Wildl. Manage. 14:115-123.

DIXON, W.J., and M.B. BROWN. 1979. BMDP biomedical computer program, p-series. University of California Press, Berkeley. 880 pp.

EARHART, C.M. 1969. The influence of soil texture on the structure, durability, and occupancy of muskrat burrows in farm ponds. Calif. Fish Game. 55:179-196.

ERRINGTON, P.L. 1937. Habitat requirements of stream-dwelling musk­rats. Trans. N. Am. Wildl. Conf. 2:411-416.

• 1940. Natural restocking of muskrat-vacant habitats. -----J:-WTTci: Manage. 4:173-185.

• 1941. Versatility in feeding and population mainten­-ance or-the muskrat. J. Wildl. Manage. 5:68-89.

• 1951. Conce1"ning fluctuations in populations of the proii!Tc- and widely distributed muskrat. Am. Nat. 85:273-292.

1963. Muskrat populations. Iowa State University ----Press:-Ames. x + 665 pp. FREUND, R.J., and D.L. DEBERTIN. 1975. Variable selectit"m and statis­

tical significance: a sampling experiment. Amp J. Agric~ Econ. 57:721-722.

GILFILLAN, M.C. 1947. Testing methods of increasing muskrat popula­tions. Ohio Div. Cons. Nat. Resour. Proj. No. W-015-R-04~ 40 pp.

HIGGINS, G.R. 1967. Yield of streams in Massachusetts. An interim re­port. Water Resour. Cent., Univ. Massachusetts, Amherst, Massachu­setts. iv + 175 pp.

HILL, R.C., T.B. FOMBY, and S.R. JOHNSON. 1977. Component selection norms for principal component regression. Corrmun. Stat. A6:309-334.

127

BROOKS AND DODGE

LEET, L.D., and S. JUDSON. 1971. Physical geology. 4th ed. Prentice~ Hall, Inc., Englewood Cliffs, New Jersey. xi + 687 pp.

MacCONNELL, W.P., and W. NIEDZWIEDZ. 1979. Remote sensing in the White River in Vermont. Photogram. Eng. Remote Sens. 45:1393-1399

MALLACH, N. 1971. Markeirungsversuche zur analyse des aktionsraums unct der ortsbewegungen des bisams (Ondatr~ zibethica L.). [(Attempts at marking for the analysis of the area OT act 1vi ty and movements of the muskrats ( Ondatra zibethi ca L.)]. Anz. Schaedl ingskd, Pflanzenschutz. SchadlingsbeKampfung. 44:129-136.

STEWART, R.E.A., and J .R. BIDER. 1974. Reproduction and survival of ditch-dwelling muskrats in southern Quebec. Can. Field-Nat. 88: 429-436.

U.S. GEOLOGICAL SURVEY. 1979. Water-data report MA-RI-78-1. Water resources data for Massachusetts and Rhode Island. Boston, Massa­chusetts. 319 pp.

WARWICK, T. 1940. A contribution to the ecology of the muskrat (On-datra zibethica) in the British Isles. Proc. Zool. Soc. LoiiQ, Ser. A. IIO:I65-201.

WHITE, K.J. 1978. A general computer program for econometric methods­SHAZAM. Econometrica 46:239-240.

WILLIS, C.E., ET AL. 1978. Multicollinearity: effects, symptoms, anct remedies. J. N. E. Agric. Econ. Coun. 7:55-61.

WILLNER, G.R., J.A. CHAPMAN, and J.R. GOLDSBERRY. :~975. A study anct review of muskrat food habits with special reference to Max·yland, Maryland Wildl. Admin. Publ. Wildl. Ecol. No. 1. 25 pp.

ZEJDA, J. 1976. On the interaction between the water vole (Arvicola terrestris) and the muskrat (Ondatra zibethica) in habitatseleC: tion. Zool. Listy 25:229-238.

12.8

I

NUTRf

co AS~

GREG.

NOEL

VERNG

ABSTf -----trap!'

Louis

197 8'

harve

4.6 c

Chang

Iitte

3 vef

three

areas

was j

deere

dec 1 i.

this

reduc

nut r i 1

press1

not a

such '

Louisi

exotic

state'

Louisi

for 6f

the 19

I

\

L-'

r

I k~-'>i~ .... J

co

Worldwide Furbearer Conference Proceedings

August 3-11, 1980 Frostburg, tv\aryland USA

VOLUME I

Edited by

Joseph A. Chapman, Ph.D. Appalachian Envirorymental Laboratory Center for Environmental and Estuarine Studies University of Maryland Frostburg, Maryland 2 7 532

Duane Pursley, M.S. Maryland Wildlife Administration Department of Natural Resources Annapolis, Maryland 2 7 401

1 q t\f