When a whole night's observation (1949 data not yet processed) at Port Aransas, on the southern coast of Texas, on the great overland route from eastern Mexico, yields in one night in April only seven birds, the recording of no birds at a station near the mouth of the Mississippi River becomes less significant.

As I have previously remarked in this paper, the new data obtained since 1946, when I last wrote on the subject of migration in the region of Gulf of Mexico, requires that I alter materially some of my previously held views. As more and more facts come to light, I may be compelled to alter them still further. For one thing, I have come to doubt seriously the rigidity of the coastal hiatus as I envisioned it in 1945. I believe instead that the scarcity of records of transient migrants on the Gulf coastal plain in fair weather is to a very large extent the result of a wide dispersion of birds in the dense cover that characterizes this general region. I now question if appreciable bird densities on the ground ever materialize anywhere except when the spa.r.s.eness of suitable habitat for resting or feeding tends to concentrate birds in one place, or when certain meteorological conditions erect a barrier in the path of an oncoming migratory flight, precipitating many birds in one place.

This retrenchment of ideas is a direct consequence of the present study, for time and again, as discussed in the case of Tampico densities, maximal nightly flights have failed to produce a visible abundance of transients on land the following day. A simple example may serve to ill.u.s.trate why. The highest one-hour density recorded in the course of this study is 21,200 birds. That means that this many birds crossed a line one mile long on the earth's surface and at right angles to the direction of flight. Let us further a.s.sume that the average flight speed of all birds comprising this flight was 30 mph.

Had the entire flight descended simultaneously, it would have been dispersed over an area one mile wide and thirty miles long, and the precipitated density on the ground would have been only 1.1 birds per acre. Moreover, if as many as ten species had been involved in the flight, this would have meant an average per species of less than one bird per nine acres. This would have failed, of course, to show appreciable concentrations to the observer in the field the following day. If, however, on the other hand, the same flight of 21,200 birds had encountered at one point a weather barrier, such as a cold-front storm, all 21,200 birds might have been precipitated in one place and the field observer would have recorded an "inundation of migrants."

This would be especially true if the locality were one with a high percentage of open fields or prairies and if the flight were mainly of woodland dwelling species, or conversely, if the locality were densely forested with few open situations and the flight consisted mainly of open-country birds. As explained on page 389, the density formula may be too conservative in its expression of actual bird densities. Even if the densities computed for birds in the air are only half as high as the actual densities in the air, the corresponding ground density of 2.2 birds per acre that results if all the birds descended simultaneously would hardly be any more impressive than the 1.1 bird per acre.

This consideration is doubtless highly modified by local circ.u.mstances, but, in general, it seems to suggest a working hypothesis that provides an explanation for many of the facts that we now have. For example, on the coast of Texas there are great expanses of terrain unattractive to such birds as warblers, vireos, tanagers, and thrushes. The precipitation there by bad weather of even a mediocre nightly flight composed of birds of the kinds mentioned would surely produce an overwhelming concentration of birds in the scattered woods and shrubs.

In spite of all that has been written about the great concentrations of transient migrants on the coast of Texas in spring, I am not convinced that they are of a different order of magnitude than those concentrations that sometimes occur along the cheniers and coastal islands of Louisiana and Mississippi. I have read over and over the highly informative accounts of Professor Williams (_loci cit._) and the seasonal summaries by Davis (1936-1940) and Williams (1941-1945).

I have conversed at length with Mrs. Jack Hagar, whom I regard as one of the leading authorities on the bird life of the Texas coast, and she has even permitted me access to her voluminous records covering a period of fifteen years residence at Rockport.

Finally, I have spent a limited amount of time myself on the Texas coast studying first-hand the situation that obtains there in order that I might be in a position to compare it with what I have learned from observations elsewhere in the region of the Gulf of Mexico, Louisiana, Florida, Yucatan, and eastern Mexico.

Although the concentrations of birds on some days near the mouth of the Mississippi River are almost incalculable, the fact remains that in Texas the densities of transient species on the ground are more consistently high from day to day. The reason for this may be simple.

As birds move up daily from Mexico overland, a certain percentage would be destined to come down at all points along the route but so dispersed in the inland forest that they might pa.s.s unnoticed.

However, that part of the same flight settling down in coastal areas, where trees are scarce, would produce visible concentrations of woodland species. With the advent of a cold-front storm, two diametrically opposite effects of the same meteorological phenomenon would tend to pile up great concentrations of migrants of two cla.s.ses--the overland and the trans-Gulf flights. During the prepolar-front weather the strong southerly (from the south) and southeasterly winds would tend to displace much of the trans-Gulf segment to the western part of the Gulf. With the shift of the winds to the north and northwest, which always occurs as the front pa.s.ses, the overland flight still in the air would tend to be banked up against the coast, and the incoming trans-Gulf flight would be confronted with a barrier, resulting in the precipitation of birds on the first available land.

These postulated conditions are duplicated in part in autumn along the Atlantic coast of the eastern United States. There, as a result of the excellent work of Allen and Peterson (1936) and Stone (1937), a similar effect has been demonstrated when northwest winds shove the south-bound flights up against the coast of New Jersey and concentrate large aggregations of migrants there.

_Interior of the United States_

Attention has been drawn already to the nature of the nightly flights at stations immediately inland from the Gulf coast, where densities decline abruptly well before midnight. I have suggested that this early drop-off is mainly a result of the small amount of terrain south of these stations from which birds may be contributed to a night's flight. At Oak Grove, Louisiana, the flight exhibited a strong directional trend with no significant aberrant components. Therefore, one may infer that a considerable part of the flight was derived from regions to the south of the station.

At Mansfield, Louisiana, thirty-eight hours of observation in April and May resulted in flight densities that are surprisingly low--much lower, in fact, than at Oak Grove. In eleven of the hours of observation no birds at all were seen. A possible explanation for these low densities lies in the fact that eastern Texas and western Louisiana, where, probably, the Mansfield flights originated, is not an especially attractive region to migrants because of the great amount of deforested and second growth pine land. Oak Grove, in contrast, is in the great Tensas-Mississippi River flood plain, characterized by an almost solid stand of deciduous forest extending over thousands of square miles in the lower Mississippi valley.

[Ill.u.s.tration: FIG. 37. Sector density representation on two nights at Rosedale, Mississippi, in 1948. The white lines are the vector resultants.]

In further contrast to the considerable flight densities and p.r.o.nounced directional trend at Oak Grove, we have the results from Rosedale, Mississippi, only seventy miles to the north and slightly to the east. At Rosedale the densities were mediocre and the flight directions were extremely divergent. Many of the nights of observation at this locality were seriously interrupted by clouds, but such counts as were made on those dates indicated little migration taking place.

On two nights, however, April 21-22 and May 20-21, visibility was almost continuous and densities were moderately high. In Figure 37 I have shown the flight directions for these two nights. The lengths of the individual sector vectors are plotted as a percentage of the total station density for each of the two nights (5,800 and 6,800 birds, respectively). Although the vector resultants show a net movement of birds to the northeast, there are important divergent components of the flights. This "round-the-compa.s.s" pattern is characteristic of stations on the edge of meteorological disturbances, as was Rosedale on April 21-22, but not on the night of May 20-21. If bats are presumed to have played a role in these latter observations, their random flights would tend to cancel out and the vector resultant would emerge as a graphic representation of the actual net trend density of the birds and its prevailing direction of flow. Although I do not believe that bats are the real reason for the diverse directional patterns at Rosedale, I can offer no alternative explanation consistent with data from other stations.

Moving northward in the valley of the Mississippi and its tributaries, we find a number of stations that yielded significantly high densities on most nights when weather conditions were favorable for migration.

Louisville and Murray, Kentucky, and Knoxville, Tennessee, each show several nights with many birds flying, but only Lawrence, Kansas, and Ottumwa, Iowa, had migrations that approach in magnitude the record station densities at Tampico. Indeed, these were the only two stations in the United States that produced flights exceeding the densities at Progreso, Yucatan. The densities at Lawrence are unique in one respect, in that they were extremely high in the month of March. Since there were very few stations in operation then, these high densities would be of little significance were it not for the fact that at no time in the course of this study from 1945 to the present have comparable densities been obtained this early in the migration period.

Examination of the "Remarks" section of the original data sheets from Lawrence show frequent mention of "duck-like" birds pa.s.sing before the moon. We may infer from these notations that a considerable part of the overhead flight was composed of ducks and other aquatic birds that normally leave the southern United States before the main body of transient species reach there. The heavy flight densities at Lawrence may likewise have contained certain Fringillidae, Motacillidae, Sylviidae, and other pa.s.serine birds that winter mainly in the southern United States and which are known to begin their return northward in March or even earlier. Observations in 1948 at Lawrence in April were hindered by clouds, and in May no studies were attempted. However, we do have at hand two excellent sets of data recorded at Lawrence on the nights of May 3-4 and May 5-6, 1947, when the density was also extremely high.

At Ottumwa, Iowa, where a splendid cooperative effort on the part of the local ornithologists resulted in forty-four hours of observation in April and May, densities were near the maximum for all stations.

Considering this fact along with results at Lawrence and other mid-western stations where cloud cover did not interfere at the critical periods of observation, we have here evidence supporting the generally held thesis that eastern Kansas, Missouri, and Iowa lie on a princ.i.p.al migratory flyway. Stations in Minnesota, Illinois, Michigan, Ma.s.sachusetts, and Ontario were either operated for only parts of one or two nights, or else clouds seriously interfered with observations, resulting in discontinuous counts. It may be hoped that future studies will include an adequate representation of stations in these states and that observations will be extensive enough to permit conclusions regarding the density and direction of migration.

Charleston, South Carolina, which does not conveniently fall in any of the geographic regions so far discussed, had, to me, a surprisingly low flight density; twenty-two hours of observation there in March, April, and May yielded a total flight density of only 3,000 birds.

This is less, for example, than the number of birds computed to have pa.s.sed Lawrence, Kansas, in one hour, or to have pa.s.sed Progreso, Yucatan, in one twenty-minute interval! Possibly observations at Charleston merely chanced to fall on nights of inexplicably low densities; further observations will be required to clear up this uncertainty.

E. MIGRATION AND METEOROLOGICAL CONDITIONS

The belief that winds affect the migration of birds is an old one. The extent to which winds do so, and the precise manner in which they operate, have not until rather recently been the subject of real investigation. With modern advances in aerodynamics and the development of the pressure-pattern system of flying in aviation, attention of ornithologists has been directed anew to the part that air currents may play in the normal migrations of birds. In America, a brief article by Bagg (1948), correlating the observed abundance of migrants in New England with the pressure pattern obtaining at the time, has been supplemented by the unpublished work of Winnifred Smith. Also Landsberg (1948) has pointed out the close correspondence between the routes of certain long-distance migrants and prevailing wind trajectories. All of this is basis for the hypothesis that most birds travel along definite air currents, riding with the wind. Since the flow of the air moves clockwise around a high pressure area and counterclockwise around a low pressure area, the birds are directed away from the "high" and toward the center of the "low." The arrival of birds in a particular area can be predicted from a study of the surrounding meteorological conditions, and the evidence in support of the hypothesis rests mainly upon the success of these predictions in terms of observations in the field.

From some points of view, this hypothesis is an attractive one. It explains how long distances involved in many migrations may be accomplished with a minimum of effort. But the ways in which winds affect migration need a.n.a.lysis on a broader scale than can be made from purely local vantage points. Studies of the problem must be implemented by data acc.u.mulated from a study of the process in action, not merely from evidence inferred from the visible results that follow it. Although several hundred stations operating simultaneously would surely yield more definite results, the telescopic observations in 1948 offer a splendid opportunity to test the theory on a continental scale.

The approach employed has been to plot on maps sector vectors and vector resultants that express the directional trends of migration in the eastern United States and the Gulf region, and to compare the data on these maps with data supplied by the U. S. Weather Bureau regarding the directions and velocities of the winds, the location of high and low pressure areas, the movement of cold and warm fronts, and the disposition of isobars or lines of equal pressure. It should be borne in mind when interpreting these vectors that they are intended to represent the directions of flight only at the proximal ends, or junction points, of the arrows. The tendency of the eye to follow a vector to its distal extremity should not be allowed to create the misapprehension that the actual flight is supposed to have continued on in a straight line to the map location occupied by the arrowhead.

A fundamental difficulty in the pressure-pattern theory of migration has no doubt already suggested itself to the reader. The difficulty to which I refer is made clear by asking two questions. How can the birds ever get where they are going if they are dependent upon the whim of the winds? How can pressure-pattern flying be reconciled with the precision birds are supposed to show in returning year after year to the same nesting area? The answer is, in part, that, if the wind is a major controlling influence on the routes birds follow, there must be a rather stable pattern of air currents prevailing from year to year.

Such a situation does in fact exist. There are maps showing wind roses at 750 and 1,500 meters above mean sea level during April and May (Stevens, 1933, figs. 13-14, 17-18). Similarly, the "Airway Meteorological Atlas for the United States" (Anonymous, 1941) gives surface wind roses for April (Chart 6) and upper wind roses at 500 and 1,000 meters above mean sea level for the combined months of March, April, and May (Charts 81 and 82). The same publication shows wind resultants at 500 and 1,000 meters above mean sea level (Charts 108 and 109). Further information permitting a description in general terms of conditions prevailing in April and May is found in the "Monthly Weather Review" covering these months (_cf._ Anonymous, 1948 _a_, Charts 6 and 8; 1948 _b_, Charts 6 and 8).

[Ill.u.s.tration: FIG. 38. Over-all sector vectors at major stations in the spring 1948. See text for explanation of system used in determining the length of vectors. For identification of stations, see Figure 34.]

[Ill.u.s.tration: FIG. 39. Over-all net trend of flight directions at stations shown in Figure 38. The arrows indicate direction only and their slants were obtained by vector a.n.a.lysis of the over-all sector densities.]

First, however, it is helpful as a starting point to consider the over-all picture created by the flight trends computed from this study. In Figure 38, the individual sector vectors are mapped for the season for all stations with sufficient data. The length of each sector vector is determined as follows: the over-all seasonal density for the station is regarded as 100 percent, and the total for the season of the densities in each individual sector is then expressed as a percentage. The results show the directional spread at each station.

In Figure 39, the direction of the over-all vector resultant, obtained from the sector vectors on the preceding map, is plotted to show the net trend at each station.

As is evident from the latter figure, the direction of the net trend at Progreso, Yucatan, is decidedly west of north (N 26 W). At Tampico this trend is west of north (N 11 W), but not nearly so much so as at Progreso. In Texas, Louisiana, Georgia, Tennessee, and Kentucky, it is decidedly east of north. In the upper Mississippi Valley and in the eastern part of the Great Plains, the flow appears to be northward or slightly west of north. At Winter Park, Florida, migration follows in general the slant of the Florida Peninsula, but, the meager data from Thomasville, Georgia, do not indicate a continuation of this trend.

It might appear, on the basis of the foregoing data, that birds migrate along or parallel to the southeast-northwest extension of the land ma.s.ses of Central America and southern Mexico. This would carry many of them west of the meridian of their ultimate goal, obliging them to turn back eastward along the lines of net trend in the Gulf states and beyond. This curved trajectory is undoubtedly one of the factors--but certainly not the only factor--contributing to the effect known as the "coastal hiatus." The question arises as to whether this northwestward trend in the southern part of the hemisphere is a consequence of birds following the land ma.s.ses or whether instead it is the result of some other natural cause such as a response to prevailing winds. I am inclined to the opinion that both factors are important. Facts pertinent to this opinion are given below.

In April and May a high pressure area prevails over the region of the Gulf of Mexico. As the season progresses, fewer and fewer cold-front storms reach the Gulf area, and as a result the high pressure area over the Gulf is more stable. Since the winds move clockwise around a "high," this gives a general northwesterly trajectory to the air currents in the vicinity of the Yucatan Peninsula. In the western area of the Gulf, the movement of the air ma.s.s is in general only slightly west of north, but in the central Gulf states and lower Mississippi Valley the trend is on the average northeasterly. In the eastern part of the Great Plains, however, the average circulation veers again slightly west of north. The over-all vector resultants of bird migration at stations in 1948, as mapped in Figure 39, correspond closely to this general pattern.

Meteorological data are available for drawing a visual comparison between the weather pattern and the fight pattern on individual nights. I have plotted the directional results of four nights of observation on the Daily Weather Maps for those dates, showing surface conditions (Figures 40, 42, 44 and 46). Each sector vector is drawn in proportion to its percentage of the corresponding nightly station density; hence the vectors at each station are on an independent scale. The vector resultants, distinguished by the large arrowheads, are all a.s.signed the same length, but the nightly and average hourly station densities are tabulated in the legends under each figure. For each map showing the directions of flight, there is on the facing page another map showing the directions of winds aloft at 2,000 and 4,000 feet above mean sea level on the same date (see Figures 41-47). The maps of the wind direction show also the velocities.

Unfortunately, since there is no way of a.n.a.lyzing the sector trends in terms of the elevations of the birds involved, we have no certain way of deciding whether to compare a given trend with the winds at 2,000, 1,000, or 0 feet. Nor do we know exactly what wind corresponds to the average or median flight level, which would otherwise be a good alt.i.tude at which to study the net trend or vector resultant.

Furthermore, the Daily Weather Map ill.u.s.trates conditions that obtained at 12:30 A. M. (CST); the winds aloft are based on observations made at 10:00 P. M. (CST); and the data on birds covers in most cases the better part of the whole night. Add to all this the fact that the flight vectors, their resultants, and the wind representations themselves are all approximations, and it becomes apparent that only the roughest sort of correlations are to be expected.

However, as will be seen from a study of the accompanying maps (Figures 40-47), the shifts in wind direction from the surface up to 4,000 feet above sea level are not p.r.o.nounced in most of the instances at issue, and such variations as do occur are usually in a clockwise direction. All in all, except for regions where frontal activity is occurring, the weather maps give a workable approximation to the average meteorological conditions on a given night.

The maps (Figures 40-47) permit, first, study of the number of instances in which the main trend of flight, as shown by the vector resultant, parallels the direction of wind at a reasonable potential mean flight elevation, and, second, comparison of the larger individual sector vectors and the wind currents at any elevation below the tenable flight ceiling--one mile.

On the whole, inspection of the trend of bird-flight and wind direction on specific nights supports the principle that the flow of migration is in general coincident with the flow of air. It might be argued that when the flow of air is toward the north, and when birds in spring are proceeding normally in that direction, no significance can be attached to the agreement of the two trends. However, the same coincidence of wind directions and bird flights seems to be maintained when the wind currents deviate markedly from a northward trajectory.

Figures 46 and 47, particularly in regard to the unusual slants of the flight vectors at Ottumwa, Knoxville, and Memphis, ill.u.s.trate that this coincidence holds even when the wind is proceeding obliquely eastward or westward. On the night of May 22-23, when a high pressure area prevailed from southern Iowa to the Atlantic coast, and the trajectory of the winds was northward, migration activity at Knoxville and Ottumwa was greatly increased and the flow of birds was again northward in the normal seasonal direction of migration.

Further study of the data shows fairly conclusively that maximum migration activity occurs in the regions of high barometric pressure and that the volume of migration is either low or negligible in regions of low pressure. The pa.s.sage of a cold-front storm may almost halt migration in spring. This was demonstrated first to me by the telescopic method at Baton Rouge, on April 12, 1946, following a strong cold front that pushed southeastward across the Gulf coastal plain and over the eastern Gulf of Mexico. The winds, as usual, shifted and became strong northerly. On this night, following the shift of the wind, only three birds were seen in seven hours of continuous observation. Three nights later, however, on April 15, when the warm air of the Gulf was again flowing from the south, I saw 104 birds through the telescope in two hours. Apropos of this consideration in the 1948 data are the nights of May 21-22 and 22-23.

[Ill.u.s.tration: FIG. 40. Comparison of flight trends and surface weather conditions on April 22-23, 1948. The meteorological data were taken from the U. S. Weather Bureau Daily Weather Map for 12:30 A. M. (CST) on April 23. The nightly station densities and the average hourly station density (shown in parentheses) are as follows:

5. Louisville: 9,100 (1,100) 6. Murray: 16,300 (2,700) 8. Stillwater: 1,900 (500) 9. Knoxville: 15,200 (1,700) 13. Oak Grove: 13,600 (1,700) 16. College Station: 13,300 (1,900) 17. Baton Rouge: 6,200 (1,000) 19. Lafayette: 2,800 (600) 21. Winter Park: 6,200 (700) 23. Tampico: 11,100 (3,700)]

[Ill.u.s.tration: FIG. 41. Winds aloft at 10:00 P. M. on April 22 (CST). Winds at 2,000 feet above mean sea level are shown in black; those at 4,000 feet, in white. Velocities are indicated by standard Beaufort Scale of Wind Force. The numbers in circles refer to the stations shown in Figure 40.]

Correction: Figures 41 and 45 were inadvertently transposed.

[Ill.u.s.tration: FIG. 42. Comparison of flight trends and surface weather conditions on April 23-24, 1948. The meteorological data were taken from the U. S. Weather Bureau Daily Weather Map for 12:30 A. M. (CST) on April 24. The nightly station densities and the average hourly station density (shown in parentheses) are as follows:

1. Albion: 1,100 (300) 2. Ottumwa: 5,500 (900) 4. Lawrence: 5,400 (1,400) 5. Louisville: 13,300 (2,700) 6. Murray: 9,800 (1,400) 8. Stillwater: 800 (100) 9. Knoxville: 8,000 (900) 10. Memphis: 7,900 (1,000) 14. Mansfield: 4,900 (1,200) 16. College Station: 700 (100) 17. Baton Rouge: 1,700 (400) 18. Pensacola: migration negligible 20. New Orleans: 1,600 (800) 21. Winter Park: 2,700 (300) 23. Tampico: 63,600 (6,300) 24. Progreso: 31,300 (3,900)]

[Ill.u.s.tration: FIG. 43. Winds aloft at 10:00 P. M. on April 23 (CST). Winds at 2,000 feet above mean sea level are shown in black; those at 4,000 feet, in white. Velocities are indicated by standard Beaufort Scale of Wind Force. The numbers in circles refer to the stations shown in Figure 42.]

[Ill.u.s.tration: FIG. 44. Comparison of flight trends and surface weather conditions on April 24-25, 1948. The meteorological data were taken from the U. S. Weather Bureau Daily Weather Map for 12:30 A. M. (CST) on April 25. The nightly station densities and the average hourly station density (shown in parentheses) are as follows:

1. Albion: migration negligible 2. Ottumwa: 4,600 (1,500) 3. Columbia: 1,400 (400) 5. Louisville: 1,700 (200) 10. Memphis: 6,600 (900) 12. Rosedale: 1,100 (100) 14. Mansfield: 1,700 (400) 18. Pensacola: migration negligible 21. Winter Park: 600 (100) 24. Progreso: 27,300 (3,000)]

[Ill.u.s.tration: FIG. 45. Winds aloft at 10:00 P. M. on April 24 (CST). Winds at 2,000 feet above mean sea level are shown in black; those at 4,000 feet, in white. Velocities are indicated by standard Beaufort Scale of Wind Force. The numbers in circles refer to the stations shown in Figure 44.]