Miller and Ralph 



Chapter 20 



Inland Habitat Relationships in California 



Table 1 Ten-day intervals and corresponding standard- 

 ization factors for seasonal variation of mean Marbled 

 Murrelet levels at three sites in northern California 



Thus for surveys conducted at the three sites from 14 

 July to 23 July, numbers of detections per survey were on 

 average 1 .59 times greater than the summer mean; surveys 

 conducted from 15 May through 24 May had numbers of 

 detections which were about equivalent to the summer mean; 

 and numbers of detections for surveys from 25 April to 4 

 May averaged about half of the summer mean. 



In applying the standardization, we made the assumption 

 that the relationship between detections at any site on a 

 given day and the mean detection levels for the summer 

 period at that site would be the same as the relationship we 

 found at the three test sites. We have compared data with 

 one site with very low activity and found the seasonal curves 

 were similar. Standardized mean detection levels were 

 calculated for all stands and stations and this measure used 

 for all analyses. 



Stand Study: Isolated Stands 



Multiple Regression 



We examined the relationship between standardized mean 

 detection levels for the stand, referred to as the dependent 

 variable, and the following independent variables: stand 

 size, Patton's index of perimeter to area (Patton 1975) which 

 was used as a measure of the edge or shape, distance from 

 salt water, density of old-growth trees, type of stand, and 

 dominant tree species. As a transformation of the standardized 

 mean detection level, we used the square root of the mean 

 for the multiple regression. 



Logistic Regression 



For each stand we summarized the detections and 

 behaviors for all surveys conducted during the study to 

 determine the status of the stand. If no murrelets were detected 



during any of the surveys, then the status was "Undetected." 

 Stands with murrelet detections were assigned a status of 

 "Present" or, if occupied behaviors (Paton, this volume; 

 Ralph and others 1 993) were observed, a status of "Occupied." 

 Using logistic regression (SAS Institute, Inc. 1991) 

 with maximum likelihood analysis of variance, we examined 

 the relationship between a selection of independent variables, 

 and status. We compared response variables Present 

 (including Occupied stands) and Undetected, and response 

 variables Occupied and Unoccupied (all stands with a status 

 of Undetected or Present). For the stands with murrelets 

 present we compared Occupied stands, with stands with a 

 status of Present. 



Park Study: Large Contiguous Stands 



Elevation and position on the landscape were estimated 

 from topographic maps to give a measure of topography for 

 each station. Landscape position was described as one of 

 five categories: (1) in the bottom of a major drainage, a 

 drainage covering a large length of the landscape and isolated 

 by parallel ridges; (2) in the bottom of a tributary (or minor) 

 drainage, a drainage flowing into a major drainage, or a 

 short, steep drainage flowing directly into the ocean; (3) on 

 top of a major ridge, a ridge running parallel to a major 

 drainage; (4) on top of a minor ridge, a ridge line that 

 originated from the major ridge and was generally 

 perpendicular to a major drainage; and (5) on a general 

 slope, a station not on a ridge nor in a drainage. 



When stations were located on slopes or ridges, it was 

 possible to detect murrelets calling in the drainages. The 

 topography within 100 m of the stations was similar to the 

 topography at the station itself. To help isolate the effects 

 of topography, we included only detections within 100 m 

 of the observer. 



Results 



Stand Study: Isolated Stands 



We identified 286 potential study stands in Del Norte, 

 Humboldt, Trinity, San Mateo, and Santa Cruz counties 

 meeting the criteria in the four size categories 2 to 20 ha (n = 

 184); 21 to 40 ha (n = 39); 41 to 100 ha (n = 35); >100 ha (n 

 = 28). We located few stands >2 1 ha, therefore, we surveyed 

 all accessible stands in those categories. From these potential 

 study stands we selected and surveyed 152 stands as follows: 

 2 to 20 ha (n = 86); 21 to 40 ha (n = 22); 41 to 100 ha (n = 

 23); >100 ha ( = 21). Due to weather conditions, three 

 stands were surveyed only three times. 



Density of the combined dominant and codominant tree 

 cover and presence of redwood trees were positively and 

 significantly (F 0>05 = 2.428, df model = 10, P = 0.0105, R 2 = 

 0.1625) related to mean murrelet detection levels in the 

 multiple regression model. Because only 16 percent of the 

 variation in the system was explained by the model, the 

 predictive ability was limited. Other variables examined 

 were not related to mean detection levels. 



210 



USDA Forest Service Gen. Tech. Rep. PSW-152. 1995. 



