Habitat Distribution and Territoriality In the Dickcissel and Red-Winged Blackbird

Results and Discussion

The Dickcissel

The dickcissel is a small territorial bird typically found in meadows, roadside edges and ditches, and oldfield habitats. It breeds from the Gulf Coast to the Canadian border and from the Western Plains to the Appalachian Mountains. The dickcissel (Spiza americana) has recently been placed on the Blue List as a threatened species. This came about as the result of research conducted by Fretwell (1977).

Zimmerman (1971) found that late oldfield habitats are particularly preferred by the dickcissel and that territoriality in the dickcissel evidently has a density - limiting function. He observed that at high densities, some birds were forced into suboptimal habitats by settled birds. Fretwell and Calver (1969) also observed the forcing of some males into less suitable habitats and also concluded that territorial behavior was limiting density in the more preferred habitats.

Density Dynamics

Assuming the findings of Zimmerman, Fretwell, and Calver to be correct, I predicted that dickcissel densities should rise to a maximum in the oldfield while prairie densities remained at a low level since the oldfield is presumably the preferred habitat. Once a maximum density is reached in the oldfield, any incoming birds should be forced into suboptimal habitat. This would be manifested by a sharp increase in prairie densities. All of this assumes that there is a steady influx of birds from somewhere outside of this system. This is a reasonable assumption since wheat and alfalfa cuttings in the area would force any birds residing in these fields to seek new habitat. This would make for an influx of birds to other areas even after migrations from the south had ceased. These predictions are summarized in the model below.

Figure 1 is a graph of the changes in observed dickcissel densities over the time span of this study in the three dickcissel study areas. Densities were consistently much higher in the oldfield than in either of the prairie plots. Density climbed sharply in the oldfield throughout the month of June while prairie densities remained virtually unchanged at a relatively low level. It would appear that the only significant increase in unburned prairie density occurred after oldfield density had reached a maximum. However, after the increase in unburned prairie density was observed, a similar increase in oldfield density occurred up to a new maximum. This makes it almost impossible to draw any conclusions as to whether any birds are being forced to settle somewhere where they wouldn't normally.

Curiously, burned prairie density remained at a stable level all through the study. Apparently, all utilizable habitat was used up at an early date so that the burned prairie was actually at a maximum density. All dickcissels on the burned prairie occupied "bottom" habitats, that is, areas which were parts of natural drainage systems and which contained tall dense growths of forbs. The short vegetation of the upland burned prairie was evidently not suitable. The fact that the bottoms filled up soon would imply that these areas have high suitability. The unburned prairie followed a similar pattern. Bottom habitats were invariably the first to be filled. In contrast to the burned prairie, however, several birds took up upland habitats on the unburned prairie after most of the bottom space was occupied. Vegetation of the upland unburned prairie was tangibly taller and denser than that of the upland burned prairie. Since bottoms were filled first and since upland unburned prairie took on birds while upland burned prairie did not, dickcissels are evidently very sensitive to the height and density (volume) of vegetation. Zimmerman (1971) discovered this in a more quantified manner.

In any event, the predicted model of density dynamics was not really satisfied. It is difficult to explain the changing of densities when one does not know for certain where the new birds are coming from. Some changes in densities were observed long after most wheat and alfalfa cuttings had been completed. It is possible that some northward migrations from Oklahoma and Texas occur well after the breeding season is under way. Evidence for a density - limiting territorial system is certainly inconclusive on the basis of density dynamics.

Suitability

In categorizing a particular habitat distribution as approaching ideal free or ideal dominance, one must observe differences in density and differences in habitat suitabilities. Differences in density are relatively simple to quantify. Observing and quantifying differences in suitability are different matters. What looks to you and I as if it might be suitable for a bird, may not necessarily be suitable from the bird's point of view. Therefore, simply recording vegetational data does not give a measure of suitability. What is needed is an index that will give a true indication of actual suitability and not reflect the observer's own biased judgment of suitability. Since suitability of a habitat "is a reflection of the average genetic contribution of resident adults to the next generation" (Fretwell and Lucas, 1969), a measure of breeding success in a habitat should be a good index of the suitability there. Fretwell (1969) used breeding success as an index of suitability in a study of field sparrow distributions.

Zimmerman (1966) has shown that the dickcissel is a polygymous species. Therefore, every female is free to choose a mate. Assuming that only males are territorial and that each female will choose the habitat that is most suitable, the most suitable habitats will end up with more females per male than less suitable habitats. Hence, the female/male sex ratio should also be a good index of suitability. This is very similar to the reasoning used by Fretwell and Calver (1969) in their study of sex ratio variation in the dickcissel. Now we are able to measure relative habitat suitabilities using either breeding success or sex ratio (or both) as suitability indexes.

Breeding Success

Breeding success can manifest itself in many ways in a population of birds. Therefore, no one measure will be sufficient basis for a conclusive statement of the breeding success of a population. I have used several components to give an over-all picture of breeding success. Table 1 lists these components and the numerical values obtained in each of the study areas. The scarcity of data from the burned prairie is due to the fact that only two nests were discovered and neither was observed through the end of the nesting term.

Table 1. Breeding Success of Dickcissels
	Oldfield	Unburned Prairie	Burned Prairie
Egg Failure	0.146	0.333	----
Fledging Success (F/N)	1.48	0.80	----
F/H	0.69	1.00	----
Parasitism (%)	58.3	87.5	100
Predation (%)	56.5	100	100
Dotted dark line indicates that data was not available. See text for explanations of the components and their calculations.

"Failed" eggs are those which survived to hatching time but failed to hatch. Egg failure was calculated as the ratio of failed eggs to the total number of eggs that survived to hatching time.

Fledging success was calculated as the ratio of total number of fledglings observed to the total number of nests observed. An F/H ratio was also calculated. It is expressed as the total number of fledglings to the total number of hatched eggs.

Parasitism (brown-headed cowbird) and predation percentages were calculated for each habitat as the percentage of parasitized and predated nests, respectively. A parasitized nest is one with at least one cowbird egg in it. A nest was scored as predated if at least one egg or one nestling was found to be missing from the nest.

Of these five components, fledging success is probably the strongest since it takes into account the amount of parasitism, predation, and egg failure. No attempt was made to rank the components, however.

Egg failure appears to clearly favor the oldfield; however the difference is not statistically significant. Fledging success also favors the oldfield, but again the difference is not significant. The F/H ratio favors the unburned prairie but evidently not significantly so. Parasitism favors the oldfield over both plots of prairie and the unburned prairie is favored slightly over burned prairie. Once again, no differences are significant. Predation was significantly lower in the oldfield than on the unburned prairie (x² = 4.55, df = 1, P < .05). The difference was the same but not significant between oldfield and burned prairie. This is due to the small sample size (two nests) from the burned prairie. It seems surprising that so many seemingly large differences exist yet so few of them are significant. A chi-square test was the best available test for these data but a chi-square test cannot be expected to detect significant differences if sample sizes are small: fewer than 10 nests were observed on either portion of the prairie.

From this data we are led to the conclusion that the oldfield is of higher suitability than the unburned prairie. There is too little data from the burned prairie to make any valid comparisons. Basing the above conclusion on only one statistically significant difference out of five components may seem a bit shaky. But if the observed differences were truly "unreal" (random) as the statistics seem to imply, then one would expect just as many of these differences to fall in favor of one habitat as the other. The fact that the oldfield is favored in three of four of these "random" differences would indicate that the differences may in fact be real. They were not detected by a statistical test because of small sample size. This, I believe, supports the conclusion made.

Sex Ratios

It is instructive to see whether observed sex ratios will support breeding success data since the female/male sex ratio is also assumed to be an index of suitability

Sex ratios were recorded as censuses were taken. A male and female were always considered mated if they were frequently seen within a few feet of each other or if the female was discovered to have a nest in the male's territory. Any copulations were also considered as proof of a pair bond (Zimmerman, 1966). A male was also considered to have at least one mate if persistent alarm calls were given when the bird was approached (Fretwell and Calver, 1969). Bachelor males were never observed to give alarm calls. Frequent traverses were made through males' territories to flush any females that were hidden.

Table 2 gives the weekly sex ratios in each study area. Sex ratios were consistently higher in the oldfield than on either prairie plot. As with breeding success data, sample sizes were too small to produce significant differences with a chi-square test. In only two censuses were there more than 15 males on the unburned prairie. Five was the maximum number of males ever observed on the burned prairie. The small amount of data from the burned prairie (5 males) would make any conclusions concerning this plot unreliable. Therefore, the discussion is limited primarily to oldfield and unburned prairie.

Table 2. Dickcissel Sex Ratios (Female/Male)
Time Period	Oldfield	Unburned Prairie	Burned Prairie
6/9-6/16	0.92	0.50	0.40
6/21-6/26	1.00	0.67	0.00
6/27-7/3	1.19	0.80	0.60
7/4-7/12	0.84	0.89	0.80
7/13-7/19	1.07	0.94	1.00
7/28-7/31	1.86	----	1.67
All Dates	1.16	0.81	0.70
All sex ratios obtained by summing total number of males and females present during each time period and calculating the ratio. Dotted dark line indicates no males present on unburned prairie for that time period.

An average sex ratio was computed for each area. The value for the oldfield proved to be significantly higher than that of the unburned prairie (t=2.430, df=9, P<.05) and burned prairie (t=1.620, df=10, P<.07). No significant difference exists between burned and unburned prairie. The oldfield, then, would have to be considered to have the edge in suitability. Thus, sex ratio data support breeding success data and comparisons of suitabilities determined by both methods agree. These findings support those of Fretwell and Calver (1969) who discovered that dickcissel sex ratios were consistently higher in clover fields than in pastures (analogous to oldfield and prairie) in an east - west transect of the dickcissel's summer range.

A plot of sex ratio vs. density is shown in figure 2. The best-fit regression line has been imposed on the scattergram. The correlation does not appear to be strong, but it is significant (t=1.888, df=13, P<.08). A similar positive correlation between sex ratio and male density was observed by Zimmerman (1971). He obtained a parabolic regression, however, in which sex ratio reached a peak at about 70 males/100 acres and then began to decrease at higher densities. The increase in sex ratio is to be expected since, in most dickcissel habitats, the volume of the vegetation increases throughout the breeding season, making male's territories more attractive to females. Zimmerman believed the decrease in sex ratio at high densities was due to the forcing of some males into less suitable habitat patches within the same field, decreasing the desirability of their territories to females. The linear regression obtained for the data of this study certainly does not imply a decrease in sex ratio at high densities. But the territories of males entering the oldfield at high densities were actually observed to contain significantly taller vegetation (see Table 4) than those of males that were already established. This would suggest that the territories of "late" males are just as suitable, if not more so, than those of "early" males. Thus, not a decrease, but an increase in sex ratio would be expected.

It will be remembered that the two types of ideal habitat distributions carried with them the assumption that habitat suitability always decreases with an increase in male density. Obviously, that assumption fails in our consideration of the dickcissel since the habitat suitability index (sex ratio) is positively, not negatively, correlated with male density. I do not believe, however, that the failure of this assumption invalidates an application of the theory to the dickcissel habitat distribution, especially since the positive correlation between suitability and density was observed in all habitats. It only implies that the mechanisms by which the final habitat distribution is achieved will be different than those outlined in the introductory theory. That is, the habitats of different suitability may not necessarily fill up in a sequential manner. This was observed in the data and discussion of density dynamics.

Habitat suitability is observed to increase with increasing density. If the ideal and free assumptions hold, all incoming birds would crowd into the best habitat (oldfield) since suitability would never decrease to a level that would equal maximum suitability in a poorer habitat. Since this was not observed, it would appear that the dickcissel distribution cannot fit the ideal free model. On the other hand, if the ideal and dominance assumptions hold, birds would fill one habitat until the density reached such a level that the apparent suitability to the senses of an incoming bird would be as low as the maximum suitability in a poorer habitat. Apparent suitability may be expected to decrease even though actual suitability is increasing, since, as Huxley (1934) observed, resistance from each settled male increases as territory size decreases (as density increases). These observations suggest, then, that the dickcissel distribution may approximate the ideal dominance distribution. This remains to be verified.

Territoriality

Territorial characteristics of the dickcissel may well provide more clues as to the type of habitat distribution exhibited by the dickcissel, and hence, to the function of its territorial behavior. The male dickcissel commonly sings from conspicuous perches at different points within or along the boundary of its established territory. The dickcissel has also been observed to engage in vigorous territorial defense, giving heated chase to any other male who should stray too closely (Schartz and Zimmerman, 1971). These phenomena were readily observed in this study.

Figure 3 shows an inverse correlation between male density and territory size in the dickcissel. A parabolic regression curve has been fitted to the points. It shows that a minimum territory size of about 0.35 acres was reached at an approximate density of 100 males/100 acres. A minimum territory size implies that an increase in density beyond the point at which this minimum size was reached will result in the utilization of less suitable habitat patches within the same field by incoming males. This is because the resident males, occupying the most suitable patches, would not allow their territories to be compressed any further. More importantly, if we assume that territories do not overlap (which is a valid assumption), a minimum territory size would impose a limit on the density which would be reached when all space was utilized. According to the data, this limit would occur at roughly three males per acre. Then any incoming males would have no choice but to settle in a different habitat. These observations seem to fit the ideal dominance model and strongly suggest that territoriality has a density - limiting function in the dickcissel.

The regression curve from Zimmerman's study (1971) has also been plotted in Figure 3. His data were collected from habitats quite similar to the ones of this study. The almost identical relationship between territory size and male densities obtained in the two studies is obvious. These results are mutually supporting. But there is a discrepancy between the minimum territory sizes implied by the two curves and the densities at which they were attained. There are two possible explanations for this. Harmeson (1974) suggests that the quality of the specific habitat studied influences the critical density at which some males are forced to occupy unsuitable habitat. Thus, on an area that is quite heterogeneous, the number of good patches could be limited and a low density of males could effectively occupy all suitable habitat. Evidently, the oldfield of this study was more homogeneous than the ones studied by Zimmerman. Also, Zimmerman mapped most males more than once per week and made weekly composite maps from the daily ones. This would result in a larger average territory than if only one map of each male were made per week, which is what I did.

It is interesting to note that the data points of Figure 3 were also fitted with an exponential curve which showed a very high degree of correlation. However, I do not believe that an exponential curve would give a proper interpretation of the data. According to the data, a definite minimum territory size was reached. An exponential curve does not have a minimum. A quadratic equation (parabola), on the other hand, automatically implies that an extremum (in this case, a minimum) exists. Therefore, a parabola - half gives a better reflection of what is really happening.

A comparison of territory size was made among habitats and also among males of different mated status (Table 3). The overall data support the inverse correlation just observed between density and territory size. Overall territory sizes on the oldfield were significantly smaller than those on both unburned (z = 9.926, P < .01) and burned (z = 5.609, P < .01) prairie. Overall oldfield density (see Figure 1) was significantly higher than both plots of prairie (x² = 70.87, df = 5, P < .01). No significant difference exists between overall territory sizes of unburned and burned prairie. But there was also no significant difference in density between these two plots.

Table 3. Territory Sizes (Mean Acres + S. E.) of Male Dickcissels
Dates	Mateless	Monogamous	Polygynous	All Males
Oldfield
6/9-6/12	0.419 ± .023 (2)	0.370 ± .044 (9)	0.965 ± .000 (1)	0.428 ± .006 (12)
6/21-6/22	0.368 ± .019 (4)	0.390 ± .026 (11)	0.586 ± .088 (4)	0.426 ± .003 (19)
6/27-6/29	0.271 ± .025 (4)	0.340 ± .026 (13)	0.346 ± .025 (9)	0.332 ± .002 (26)
7/4-7/5	0.250 ± .021 (10)	0.371 ± .045 (11)	0.504 ± .052 (4)	0.344 ± .003 (25)
7/13-7/15	0.307 ± .017 (4)	0.371 ± .041 (18)	0.390 ± .035 (5)	0.365 ± .003 (27)
7/28-7/29	--------	0.374 ± .069 (6)	0.434 ± .048 (14)	0.416 ± .039 (20)
All Dates	0.297 ± .016 (24)	0.368 ± .017 (68)	0.445 ± .029 (37)	0.377 ± .013 (129)
Unburned Prairie
6/16	1.232 ± .079 (2)	1.161 ± .188 (3)	--------	1.189 ± .108 (5)
6/24-6/26	1.210 ± .139 (3)	1.029 ± .186 (6)	--------	1.089 ± .136 (9)
7/1-7/3	0.614 ± .132 (2)	1.086 ± .173 (6)	0.686 ± .000 (1)	1.012 ± .144 (9)
7/11-7/12	0.918 ± .071 (4)	0.975 ± .140 (10)	0.931 ± .413 (3)	0.954 ± .106 (17)
7/18-7/19	0.617 ± .175 (2)	0.846 ± .196 (7)	1.421 ± .157 (3)	0.952 ± .149 (12)
All Dates	0.994 + .103 (13)	0.995 ± .083 (32)	1.106 ± .208 (7)	1.009 ± .062 (52)
Burned Prairie
6/16	1.324 ± .386 (3)	0.685 ± .045 (2)	--------	1.068 ± .271 (5)
6/24	1.289 ± .311 (4)	--------	--------	1.289 ± .311 (4)
7/1	0.997 ± .315 (2)	1.713 ± .637 (3)	--------	1.426 ± .432 (5)
7/10	1.588 ± .000 (1)	1.788 ± .615 (4)	--------	1.748 ± .493 (5)
7/17	--------	1.373 ± .309 (5)	--------	1.373 ± .309 (5)
All Dates	1.271 ± .198 (10)	1.466 ± .277 (14)	--------	1.385 ± .179 (24)
Numbers in parentheses are sample sizes. Dotted dark line indicates no males of that mated status present.

On the oldfield, it would appear that there is a relation between territory size and the presence of a mate. Mateless males were seen to have smaller territories than both monogamous (z = 3.050, P < .01) and polygymous (z = 4.442, P < .0l) males. The difference between monogamous and polygynous males is not significant. Zimmerman (1966) and Harmeson (1974) obtained similar results. Neither plot of prairie produced any significant differences in territory sizes as they relate to male mated status. The inconsistency of these findings makes it difficult to make any generalizations. At least in the oldfield it would appear that females are indeed positively responsive to the volume of vegetation. Since no difference exists between territory sizes of monogamous and polygymous males, it must be the vegetation within each territory that attracts the female. The smaller size of bachelor male territories lessens the probability that they will contain vegetation attractive to a female.

Habitat Selection

To gain further insight into habitat selection in the dickcissel, vegetation heights were recorded at each grid marker in the oldfield on July 14 during one of the weekly censuses. Table 4 shows how the measurements were broken down. If vegetation height is taken as a crude index of the amount of cover afforded by the vegetation, one would expect utilized vegetation to be higher than non-utilized vegetation since dickcissels respond positively to vegetational volume. This was observed to be the case as utilized vegetation averaged significantly taller than non-utilized (z = 2.205, P < .05). No differences showed up, however, among territories with different numbers of females. Therefore, there is evidently no correlation between number of mates acquired by a male and the height of vegetation within his territory. But this is in conflict with the inferences drawn from territory size -- mated status data, since there was also no apparent correlation between number of mates and size of territory. Percentage of vegetational cover could differ substantially from what would be indicated by vegetation height alone. Since the two sets of data conflict, it would appear that this is the case. This makes it unsafe to draw any conclusions about habitat selection based on a comparison of vegetation heights.

Table 4. July 14 Oldfield Vegetation Heights (Mean cm. + S. E.)
Territory	Height
Utilized vegetation	98.28 ± 3.92
Non-utilized vegetation	85.88 ± 4.03
Mateless male territories	97.93 ± 7.12
Monogamous male territories	96.64 ± 4.65
Polygynous male territories	95.47 ± 10.45
Early male territories	94.67 ± 4.52
Late male territories	109.95 ± 7.22
Vegetation heights are categorized as they pertain to male dickcissel territories.

Vegetation heights were also classified as falling into the territories of "late" or "early" males' territories. A "late" male is any male which arrived and established a territory on the oldfield after June 28, the date at which a near maximum density and near - minimum territory size was reached (see Figure 1 and Figure 3). Earlier I concluded on the basis of territory size vs. density that males arriving after a minimum territory size had been reached would be forced into less suitable habitat. Therefore, "late" males should have shorter vegetation within their territories. The data certainly do not show this. In fact, late males occupied significantly taller vegetation than early males (z=1.793, P<.05). But it was also observed that of the eight late males which entered the oldfield, six took up territories in the exact places that had been just vacated by males that had been on the field for several weeks. One of the other two late males occupied the shortest vegetation of all but one of the males on the oldfield. In light of this, it is not surprising that late males occupied vegetation that was at least as tall as that of early males. So this vegetation data does not discredit the conclusions made earlier.

Conclusions

In concluding the discussion of the dickcissel, the results must be brought together. The burned prairie has been excluded from most of the discussion since sample sizes and data accumulations from this plot were too small to allow us to conclude anything about the burned prairie with any degree of confidence. In view of the insignificant differences between burned and unburned prairie densities, and in the absence of any glaring differences in suitability, one is led to believe that there is really little difference between burned and unburned prairie as far as the dickcissel is concerned. Of course, there are differences in upland areas of both plots as was pointed out earlier, but these are not made manifest by the data since dickcissels did not occupy any real upland areas of the burned prairie. Thus, for the purposes of discussion, conclusions made about the unburned prairie will be assumed to hold true also for the burned prairie, at least for this study.

The density dynamics hinted at a density - limiting territorial system, but provided no conclusive results as to the function of territoriality or as to the type of habitat distribution in the dickcissel.

Suitability on the basis of both sex ratios and breeding success was concluded to be higher in the oldfield than in the unburned prairie. To apply the theory developed in the introduction, we must compare this with densities. Density was observed to be much higher in the oldfield than in the unburned prairie. Thus, there is a positive correlation between suitability and density between the two habitats. This means that the habitat distribution of the dickcissel most closely approximates the ideal dominance distribution and territoriality in the dickcissel is concluded to have a density - limiting function in the absence of evidence of any other density - limiting factor. The data on sex ratios vs. density showed that the habitat distribution most probably could not be ideal free.

The data on territoriality, specifically that data comparing territory size with density, was also concluded to give strong support to the density - limiting territorial hypothesis.

It would be dangerous to generalize the conclusions of such a local study to all dickcissel populations. But the regional studies of Fretwell and Calver (1969) plus the results of Zimmerman's (1971) and Harmeson's (1974) studies all support the conclusions made and suggest that the ideal dominance distribution and density - limiting territorial system may be well - established phenomena for most dickcissel populations.

Previous Section -- Methods
Return to Contents
Next Section -- The Red-Winged Blackbird