9307 ‎(2)‎

• UCREFRP Catalog Number: 9307
• Author: Germaine, S. S.
• Title: Relationships of Birds, Lizards, and Nocturnal Rodents to Their Habitat in the Greater Tucson Area, Arizona.
• USFW Year: 1995.
• USFW - Doc Type: Phoenix, AZ.
• Copyright Material: NO

RELATIONSHn'S OF BIRDS, LIZARDS, AND NOCTURNAL RODENTS TO THEIR HABITAT IN TUCSON, ARIZONA
20
Yq
O 10
U
15
ilrJ
House mouse ®Desert pocket mouse O Barley's pocket mouse Merriam's kangaroo ret
® VVhke-tlnoated wood rat richness ~ abundance
1il)C:
LD
l,rJl;
Figure 6. Nocturnal rodent abundances in 4 housing densities in Tucson in 1994: high density, high density control, low
density, and low density control. (Total abundance square root transformed for graphing purposes only).
the non-native bird group was positively
correlated with percent area of: apartments and
small businesses; paved or graded land; urban-open
non-native vegetation cover; and distance from:
mainlands; > 1 ha+ patches; and undisturbed
washes. Non-native bird species richness was
negatively correlated with percent area upper
Sonoran vegetation, and undisturbed washes.
The best predictive equation for non-native
bird species richness (NNATR) included house
density (HDEI~, percent area: paved or graded
(PCTP), non-native urban-open vegetation
(LJROE), and upper Sonoran vegetation (USV):
NNATR = 0.545 + 0.196(I-IDEN) +
0.026(PCTP) + 0.009(LTROE) -
0.002 (LJSV) a
This model (F = 125.93; df = 6, 314; P <0.001),
had a multiple correlation coefficient of 0.84, and
explained 71% of the variation in NNATR. The
MSPR (0.745) was smaller than the MSE (0.763),
indicating that the error mean square for this
model gives an unbiased and appropriate
indication of its predictive ability. In addition,
there was no difference in slopes between models
generated with predictive and testing data sets (P
= 0.450).
Native breeding bird species richness
(BREEDR) was positively correlated with percent
area upper Sonoran vegetation and area of
undisturbed washes, and negatively correlated
with house density and percent area of: paved or
graded land; urban open non-native vegetation
cover; and all 3 distance measures. The best
predictive equation included house density
(HDEN), percent upper Sonoran vegetation
(LTSV), and distance from undisturbed washes
(DISR):
BREEDR = 10.803 - 0.466(HDEN) +
0.023(LJS~ - 0.001(DISR).
This model (F = 133.79; df = 3, 317; P <0.001)
had a multiple correlation coefficient of 0.75, and
explained 56% of the variation in BREEDR. The
MSPR (3.571) was larger than the MSE (2.153).
Therefore, the MSPR value should be referred to
STEPHEN S GERMAINE 1995 ARIZONA GAME fr FISH DEPARTMENT, TECH. REP. 20 25