A report in the Isle of Wight County Press on 6 th May 1994 stated:
Casts of large paw prints taken in a garden at St. John's Wood Road, Ryde, were sent by naturalist Martin Trippett to Dr. Karl Shucker for identification. The prints measured 4 by 4.5 inches and showed no claws. Karl identified the prints as canine rather then feline, and Ryde police were able to confirm that a Great Dane had been reported missing in the area on the previous day. Which proves yet again that the presence or absence of claws cannot be used as a definite indicator of their origin.
The following is an excellent paper which discussed just that problem.
Department of Wildlife and Fisheries Biology
University of California, Davis
(HTML version of: Smallwood, K.S. and E. L. Fitzhugh. (1989) Differentiating mountain lion and dog tracks. pp. 58-63 in R.H. Smith, ed., Proceedings, 3rd mountain lion workshop, December 6-8, 1998, Prescott, Arizona.)
Mountain lions (Felis concolor) are cryptic and occur sparsely, so sportsmen, managers, and researchers use their tracks to determine presence and relative abundance (Fitzhugh and Gorenzel 1985, Kutilek et al. 1983, Shaw 1988). Yet, many people, including trained wildlife biologists, have difficulty differentiating between lion and dog tracks (Belden, 1978). This is perfectly understandable as dog tracks vary greatly in shape and size and many are similar to lion tracks. In California, at least 9.5% of volunteers conducting a statewide track survey could not tell them apart (Smallwood and Fitzhugh, unpublished data), even though most volunteer teams included a wildlife biologist. Thus, people often question the use of tracks in mountain lion research and management. Tracking can be an important and inexpensive tool for lion study, but it must be used accurately. This paper examines currently used dog and lion track discriminators, compares them to our observations from the field, and presents the first effort to apply multivariate techniques to distinguish dog from lion tracks. This paper is relevant only for tracks from dust or firm mud because we have little experience with tracking in snow and we have observed serious track distortion in soft mud. The result is a reliable field key and office procedure.
Possible Causes for Misidentification
Illustrative and descriptive errors depicting mountain lion tracks in publications may contribute to misidentifications. A few traits of tracks we observed in the field are consistently misrepresented in publications (except Belden 1978, Downing 1979, and Downing and Fifield 1986). Our field experience was in California, but lions from other states may differ genetically and morphologically, thus they may leave tracks that look different. However, the same method we used for differentiation can be applied elsewhere to overcome possible geographical variations.
Current Techniques Used for Lion and Dog Track Discrimination
Current literature includes the following differentiators between lion and dog tracks:
We applied the traits presented in the literature as discriminators between lion and dog tracks to tracks we collected during our previous 5 years of field work. We evaluated the reliability of each of these traits in the order they were presented in the introduction, as well as for three other traits that we felt might be useful.
We traced the tracks of 19 different dogs and 48 different mountain lions onto acetate sheets after Panwar (1979). These tracks were used in a multiple group discriminant analyses to determine which of the traits presented in Figure 4 may discriminate best. Modified from Belden (1978), we used the ratio of the second toe to the heel pad (A/B). The distance between the middle toes is represented by C. The angle of the long axis of the outer toes with respect to each other was measured by first drawing a line through each of the outer toes in the direction the toes point, and then subtracting the angle E from angle D. Angles D and E were formed by the intersection of lines D and E through the base line G. The base line G was drawn tangent to the most posterior aspects of the outer 2 heel lobes. F represents the average distances of the middle toes from the front of the heel pad (a partial measure for the degree of track elongation). We further divided this average by the width of the heel pad B to normalize the values. H represents the presence or absence of claw marks, but was not tried in the analysis because the results were predictable. The concavity/convexity of the heel pad's posterior aspect was indexed by the degree and direction of discrepancy of the middle lobe's posterior aspect with the base line G. It could be concave (see track on the right), straight, or convex (see track on the left). An analysis was conducted with all tracks included, one was conducted on front tracks only, and another was conducted on rear tracks only to determine if any differences exist between front and rear track discrimination. Multiple Group discriminant analysis is a statistical technique that uses known cases to develop linear combinations of variables to predict group membership of unknown cases (Norusis, 1985). Figures 5, 6, 7, and 8 may help illustrate in an intuitive manner how the most discriminating combinations of variables are arrived at. The most discriminating variables best separate the two group distributions, and, as variables are combined, their discriminating ability is cumulated. The effectiveness of any chosen linear combination of variables is determined by the proportion of known cases correctly predicted to belong to their respective groups (dog or lion). A more rigorous description of this technique is presented in a paper showing how to identify individual mountain lions by their tracks (Smallwood and Fitzhugh in preparation).
Of the traits we tried in the analysis, the angle of the long axis of the outer toes with respect to each other best discriminated dog and lion tracks, and was indifferent to the front and rear track distinction (Table 1). The ratio of the widths of the second toe to the heel pad was also a good discriminator, but more so for front tracks than rear ones. The distance between the middle toes was a fair discriminator, but again, more so for the front tracks than rear. The distance between the middle toes and the heel pad discriminated lion and dog tracks fairly well for the rear tracks but not so well for the front tracks.
The effectiveness of multiple variables in our discriminant analyses never improved enough beyond that of the angle of the long axis of the outer toes to justify the use of more variables in the field. In fact, the second toe to heel pad ratio suppressed the effectiveness of the angle of the outer toes when they were combined in an analysis.
From this exploratory analysis, we were able to consider one new discriminating variable between lion and dog tracks, as well as another derived from Belden (1978). The effectiveness of these measurements may decrease as our sample size of dog tracks increases, but the high effectiveness in this preliminary analysis is encouraging. These variables proved very useful when combining their effectiveness with some of the variables mentioned in Part 1 to develop the following track classification key.
The following key is for distinguishing between dog and mountain lion tracks that are similar. Before using the key, screen out other species and obvious dog tracks. (Our smallest Adult lion track had a heel width of 37 mm., and lions always possess 3 lobes on the rear of the heel pad.)
|1||Heel pad concave at rear, two lobes||Dog|
|1||Heel pad with 3 lobes, shape variable||Go to 2|
|2||Front of heel pad squared or concave||Mt. lion|
|2||Front of heel pad rounded||Goto 3|
|3||Claw marks present||Go to 4|
|3||Claw marks absent||Go to 5|
|4||Claws knife-like, very narrow||Mt. lion|
|5||Angle D minus angle E = # 20°. 100% of Mt. lion, + 6% of dogs||Go to 6|
|5||Angle D minus angle E = $ 21°||94% of dogs|
|6||Check all interdependent factors to bring accuracy to 99%|
In this key, we used the best discriminating traits identified from the discriminant analyses as well as some easy-to-recognize traits we did not need to test (or could not) in the analyses. We did not observe the knife-like claw marks that identify mountain lions in step 4 of the key, but it seems reasonable that if claw marks do appear in a lion track, they should be very thin because their claws are narrow and sharp. Field biologists can considerably increase the reliability of their track identifications if, in addition to using the key, they consider the ratio of the widths of the second toe to the heel pad, whether or not there is a leading toe, the shape of the toes, the presence or absence of a mound of soil between the heel pad and the toes, the track pattern, and the travel behavior. With discriminant analysis it may be possible to identify additional traits that discriminate dog and lion tracks well.
Acknowledgments.--We thank Rob Gross and Joan Young for collecting dog tracks.
Figure 1: The typical shapes of "heels" of dog and lion tracks. The heel lobes in lion tracks are more equal in size and shape with more distinct grooves between them.
Figure 2: Typical dog and lion track patterns that we have found in the field. The arrow in center of frames is direction of travel.
Figure 3: An example of the varieties of heel pad shapes and sizes we have seen in dog tracks. The full track is presented for reference.
Figure 4: Traits used in the discriminant analyses to identify dog and lion tracks.
Figure 5: Distribution of measurements for the angle of the long axis of the outer toes for mountain lions and dogs.
Figure 6: Distribution of measurements for distance between middle toes of mountain lions and dogs.
Figure 7: Distribution of measurements for ratio of widths of second toe to the heel of mountain lion and dogs.
Figure 8: Distribution of index values for shape of heel pad base of mountain lions and dogs.
The following information was not included in the publication. Copyright by E.L. Fitzhugh 1998.
Table 1: Proportions of tracks correctly classified (and number misclassified) by different track traits from 19 dogs and 48 mountain lions in multiple group discriminant analysis. Variable labels are abbreviated: The angle of the long axis of the outer toes to each other = 'toe angles'; the ratio of the widths of the second toe to the heel pad = 'toe:heel ratio'; the average distance of toes 2 and 3 and the anterior aspect of the heel pad = "reach of toes"; the distance between toes 2 and 3 = 'spread of toes'; and, the shape of the posterior border of the heel pad = 'heel pad shape'.
|Trait||Dog(n = 63)||Lion(n = 161)||Total(n = 224)|
|Toe angles||94.0 (4)||100.0 (0)||98.3 (4)|
|Toe:heel ratio||74.6 (16)||92.5 (12)||87.2 (29)|
|Reach of toes||58.2 (26)||79.5 (33)||73.3 (60)|
|Spread of toes||88.1 (7)||75.8 (39)||79.4 (46)|
|Heel pad shape||-||-||-|
|Toe angles + toe:heel ratio||92.5 (5)||100.0 (0)||97.8 (5)|
|All 5 traits combined||97.0 (2)||100.0 (0)||99.1 (2)|
|Trait||Dog(n = 30)||Lion(n = 69)||Total(n = 99)|
|Toe angles||93.3 (2)||100.0 (0)||98.0 (2)|
|Toe:heel ratio||90.0 (3)||98.5 (1)||95.9 (4)|
|Reach of toes||66.7 (10)||75.4 (17)||72.7 (27)|
|Spread of toes||90.0 (3)||78.3 (15)||81.8 (28)|
|Heel pad shape||-||-||-|
|All 5 traits combined||100.0 (0)||100.0 (0)||100.0 (0)|
|Trait||Dog(n = 33)||Lion(n = 92)||Total(n = 125)|
|Toe angles||94.0 (2)||100.0 (0)||98.4 (2)|
|Toe:heel ratio||78.8 (7)||93.5 (6)||89.6 (13)|
|Reach of toes||57.6 (14)||85.0 (14)||77.6 (28)|
|Spread of toes||88.0 (4)||63.0 (34)||69.6 (38)|
|Heel pad shape||-||-||-|
|All 5 traits combined||94.0 (2)||100.0 (0)||98.4 (2)|
|Return to index||Return to Scottish Big Cats||Return to Identification|